~T5.6, ISO 800, 1/50th sec shutter - 2560 x 1072px version | R3D download

The Samyang 35mm T1.5 is here. It’s a cineivsed version of the popular, low cost, manual focus and manual aperture DSLR lens. The cine version is manufactured by Samyang in Korea but is also rebadged and sold under the Rokinon brand name. The DSLR version is also sold under the Bower and ProOptic names.

It features the same optics as the DSLR lens, but has a few modifications that make it cine friendly. It comes fitted with gears on the focus and aperture rings. It also features a de-clicked aperture that is marked in T-Stops rather than F-Stops.

Other Reviews

There are plenty of other reviews that go into detail regarding build quality, breathing etc so I’m not going to go into that. This review / test is mainly focused with how the lens performs on an Epic from an imaging point of view.

Images

The images in this post were all shot on a Red Epic at 5k. They have been lightly graded in Davinci Resolve. All the images can be enlarged and viewed at 2560x1350px – half of the Epic’s native 5120x2700px resolution at 1.9:1 aspect ratio. There’s a few that are 2560x1072px – 2.39:1 aspect ratio. Some also have full 1:1 resolution centre crops. No sharpening has been used on any of the images.

R3D Snaphots

I’ve provided links to R3D snapshots of each image for closer examination and so you can view them without any grading.

You’ll notice some have a green/blue cast, these have been shot using a Schneider Platinum IRND 1.2 or 1.5 filter. The cast can be fixed with the colour temp and tint controls.

All test images used in the comparisons are shot at ISO 800 with a 3:1 compression ratio. The non comparison images are shot at various ISOs and usually at 8:1 compression.

T4, ISO 320, 1/50th sec shutter - 2560 x 1072px version | R3D download

T4, ISO 500, 1/50th sec shutter - 2560 x 1072px version | R3D download

~T5.6, ISO 800, 1/50th sec shutter - 2560 x 1072px version | R3D download

Testing Methodology

The Epic was set to shoot 5k, 24fps, ISO 800 at 3:1 compression. No filters of any variety were used. The T-stop was set and and then the exposure was dialed in by varying the shutter. Exposure was set by using false colour.

Due to the strong OLPF filter, the Epic is never fully sharp at 100% zoom. So the 1:1 images can only really be compared to images from other Epics, Scarlets and similar motion cameras. Compared to images from DSLRs, they’re going to look quite soft.

All images are derived using a full quality debayer.

Test Scene 1 – Morrison St

This is a relatively high contrast scene that pushed the lens. The buildings in the centre are exposed just below clipping. The shadows under the trees are close to the noise floor.

Focus is set on the centre blue building.

Test Scene 1 – Morrison St

Below are centre 1:1 crops showing the focus area. This jpg was exported at 95% compression so as to minimise compression artefacts.

1:1 pixel crops

At the centre, the T1.5 crop seems to look softer than other shots I’ve taken at the the same T-stop. Mainly cause of the bright building in the background that seems to be causing the lens to bloom. Looking at other lower contrast areas in the T1.5 1:1 image, the lens seems to fair better. But overall at T1.5 the lens is quite soft and smeary.

The softness starts to clear at around T2 and by T2.8 it’s starting to look pretty good. At 1080p/2k or less the lens is quite usable at T2.

By T4 the lens is quite sharp. I’m struggling to see any difference in sharpness beyond T4.

Here are 2.5k images, pixel for pixel centre crops and R3D downloads of the scene.

T1.5 | 2560 x 1350px | 1:1 centre crop | R3D download
T2 | 2560 x 1350px | 1:1 centre crop | R3D download
T2.8 | 2560 x 1350px | 1:1 centre crop | R3D download
T4 | 2560 x 1350px | 1:1 centre crop | R3D download
T5.6 | 2560 x 1350px | 1:1 centre crop | R3D download
T8 | 2560 x 1350px | 1:1 centre crop | R3D download
T11 | 2560 x 1350px | 1:1 centre crop | R3D download

Test Scene 2 – Yachts

There’s a lot of detail in this shot. Probably pushing Redcode 3:1 more than the lens! It does have less contrast than the previous shot.

Focus is set on the for sale sign area.

Test Scene 2 – Yachts

1:1 pixel crops

The performance at T1.5 seems to be a bit better than in Test 1. Once again the softness improves at around T2 – T2.8. No real discernable differences beyond T4.

Here are 2.5k images, pixel for pixel centre crops and R3D downloads of the scene.

T1.5 | 2560 x 1350px | 1:1 centre crop | R3D download
T2 | 2560 x 1350px | 1:1 centre crop | R3D download
T2.8 | 2560 x 1350px | 1:1 centre crop | R3D download
T4 | 2560 x 1350px | 1:1 centre crop | R3D download
T5.6 | 2560 x 1350px | 1:1 centre crop | R3D download
T8 | 2560 x 1350px | 1:1 centre crop | R3D download
T11 | 2560 x 1350px | 1:1 centre crop | R3D download
T16 | 2560 x 1350px | 1:1 centre crop | R3D download

Test Scene 3 – Girl on street

Here is a closer shot that gives you a feel for the bokeh. Don’t judge the cinematography on this – it was a quick shot and the bounce coming from the house on the camera left is quite ugly.

There’s a slight flare from the sun hitting the lens. It’s quite subtle and doesn’t seem to be affecting the sharpness.

The focus is on the sunglasses frame, camera right. Focus is about 2 metres (6.5 feet) away from the camera.

Test Scene 3 – Girl on street

This test showed similar performance to the previous two tests. At T1.5 the dark vignetting of the corners is probably a bit more visible than with the other two tests. As with the softness, it starts to clear at around T2.

Here are 2.5k images, pixel for pixel centre crops and R3D downloads of the scene.

T1.5 | 2560 x 1350px | 1:1 centre crop | R3D download
T2 | 2560 x 1350px | 1:1 centre crop | R3D download
T2.8 | 2560 x 1350px | 1:1 centre crop | R3D download
T4 | 2560 x 1350px | 1:1 centre crop | R3D download
T5.6 | 2560 x 1350px | 1:1 centre crop | R3D download
T8 | 2560 x 1350px | 1:1 centre crop | R3D download
T11 | 2560 x 1350px | 1:1 centre crop | R3D download
T16 | 2560 x 1350px | 1:1 centre crop | R3D download

Conclusion

Most DSLR lenses aren’t perfectly sharp at f1.4. So there are no surprise that this lens is a bit soft wide open. Still, in the right circumstances T1.5 can be quite useful. It can look perfectly ok in low contrast scenes. And it can get you out of trouble if you’re willing to live with a bit of softness. At T2 the performance is good, more than usable at 1080p. At T2.8+ the lens really starts to sharpen up.

If you’re on a budget, actually even if you’re not on a budget this lens is a good buy. When you take a look at the price - US$549, this lens is a no brainer.

The post Samyang / Rokinon 35mm T1.5 Lens Test appeared first on JuanMelara.com.au.

I’ve been closely following the development of the Blackmagic Cinema Camera and have been looking forward to grading some of it’s raw files. The other day I stumbled onto this post on Fstoppers.com which links to some raw sample footage.

The video above and frame grabs below were graded in Davinci Resolve from 2400×1350 raw DNG files.

My first impression from this and other BMCC footage I’ve graded is that it compares quite favourably with Epic and Scarlet footage when in the grading suite. Sure there isn’t as much resolution but the dynamic range and colour science seems to be fairly similar.

In someways I find the BMCC produces slightly more organic footage. This may be due to the lower resolution but It may also be due to the size and quality of the noise present in some of the footage – it looks very similar to film grain.

Davinci Resolve node setup

Above is the Resolve node structure used for these two grades. The first node is a general image balance where the highlights are recovered, shadows brought up etc. Basically anything that can’t be achieved through the BMCC raw processing tab.

After that are two parallel nodes that immediately reference the first image node. One of these nodes is where all skin tweaks are done. It’s much easier to qualify skin prior to applying any creative grading, which is why it’s directly after the first node. Any other nodes that need hue based qualifications are usually placed in parallel here.

Following the parallel nodes is the shadow tint node, where the blacks are lifted and tinted. Then the high tint node where I introduce the majority of the image colour cast. Then finally is a node where I desaturate and tint just the brightest highlights to remove a bit of the effect from the previous node. This helps create a more natural looking image even under very stylised grades.

Click the images below for full size 1080p files.

More BMCC colour grading tests to come…

BMCC Grade A01

BMCC Grade A02

BMCC Grade A02 full size crop

BMCC Grade B01

BMCC Grade B03

BMCC Grade B03 fullsize crop

The post Blackmagic Design Cinema Camera Colour Grades appeared first on JuanMelara.com.au.

Ok so it’s not really recovering highlights, cause if they’re blown out, they’re gone. It’s more bringing near blown out highlights back into a useful visible range.

A few people on bmcuser and reduser have been having problems with this, so I’ve put together this post. This method is probably the easiest I’ve found and works with all footage – DSLR, BMCC, Epic, Scarlet etc. It doesn’t matter if it’s raw or not.

Blackmagic Cinema Camera

This first example is from the Blackmagic Cinema Camera. This footage is shot during scene setup by bmcuser Note Suwanchote. You can ind the original bmcuser thread here.

First thing to do is to make sure your Color Space and Gamma are set to BMD Film and your 3D Output Lookup Table is set to Blackmagic Cinema Camera Film to Rec709. This ensures you’re using all the available dynamic range, and nothing is crushed by grading Rec709 gamma encoded source material.

UPDATE: I’m finding that I’m actually getting better results by using the Arri Alexa LogC to Rec709 as the Output LUT. But test this yourself…

Next switch your colour wheels to Log mode. You’re then able to use the Highlight master wheel to bring back the highlights. By default the High Range is set too high, it only really effects the brightest highlights. Drop it to 0.333 or there abouts depending on the image.

Here is the original image.

Lowering the highlights gives you the image below. The highlights are tinted pink because the blue channel has clipped. This is made more obvious by having the Highlight Recovery box ticked since its generating the missing highlight info from the un-clipped channels.

Notice how using the Highlight wheel also affects the foliage highlights. I actually dont mind this since it creates a more natural looking image compared to a more targeted selection that only lowers the highs in the sky/clouds. Normally I wouldn’t lower the highs this much, so it won’t usually be this obvious.

Desaturate the highlights using a serial node with a Luma qualification that only affects the highs. On this same node I’ve also bumped the contrast to create more definition in the clouds. I’ve also tinted the highs a fraction blue.

Epic / Scarlet

Here is another example using Epic footage. It’s not as drastic as the example above, but usually Red footage looks pretty good if the ISO/FLUT is set properly.

It’s exactly the same process as BMCC footage, although usually you dont need a node to desaturate the highs after you’ve taken them down.

The setup for this is RedColor3 / REDlog Film. And a film emulation LUT as the 3D Output Lookup Table.

And here is the finished shot.

The post Recovering Highlights with Davinci Resolve appeared first on JuanMelara.com.au.

Reduser member Philipe Ratton posted some awesome R3Ds from his latest short film currently in production. I couldn’t resist grading these. You can find the original Reduser post with the R3Ds here.

Grading is super easy when you start with quality, well shot source material. That was definitely the case this time! There are so many possibilities with this footage. It truly can be pushed and pulled in any direction.

Resolve node structure and setup

Below is the Resolve node structure used for all these grades. You can see what effect each node has on the image in the video above.

Usually I also have a sharpen node right at the end, which I disable prior to outputting the footage. It’s really only for preview purposes since I prefer the sharpen tools in After Effects.

For some of these grades I’ve also used Resolve’s built in noise reduction. I find it helps clean up secondary qualifications. NR is always applied on the first or second node so that everything downstream benefits from a cleaner image. My settings are usually NR Threshold: 1.200, NR Blend 0.2-0.5, NR Radius: 1 for final output, 4 during grading.

Setup on this is the standard REDcolor3 / REDlog Film but this time its being graded through a film print emulation LUT. More on film print emulation LUTs to come in future posts…

The first grade is the standard teal / orange look that is popular in feature films these days. The second grade has more of a sickly green bias. And the third keeps things looking relatively neutral whilst still retaining some moodiness.

Click images for full size 1080p versions.

The post Scarlet + Davinci Resolve Colour Grading Breakdowns appeared first on JuanMelara.com.au.

BMCC + Davinci Resolve Colour Grading Breakdown from Juan Melara on Vimeo.

BMCuser.com member Kholi posted up some Blackmagic Cinema Camera raw DNGs the other day. So I thought I’d have another go at grading BMCC footage. Find the DNGs in this thread here.

The BMCC just impresses me more and more each time I work with the footage. I’m yet to shoot with it, but from a post production/grading point of view it’s a pretty amazing camera. That said, I’m yet to grade badly shot footage!

The grades

For the first grade I wanted something that looked relatively neutral, whilst still looking more polished/graded than the source footage. I still think I might have pushed it too far. The second grade is my version of Kholi’s “Eerie green” grade found in his vimeo video here. View the video above to see these grades in motion and to see a breakdown of the “Eerie green” grade.

Grading Skin

When grading I always try hard to keep the skin from looking plastic as is seen in a lot of digitally shot material. Especially feature films that have been pushed through film output emulation LUTs. You don’t have to browse Blu-ray.com or Apple Trailers for too long to stumble upon a lot of these films.

To avoid plastic looking skin I do two things:

• 1. I make sure there is colour variation within the skin tone.
• 2. I respect the specular highlights.

1. To add (or keep) variation in the skin tone I qualify the skin, and then use the colour wheels to tint the skin. I usually add blue/purple in the shadows. The midtones usually get pushed in the opposite direction that the Midtone wheel pushed in my Mid/Low Tint node. The highs are usually pushed a fraction yellow and in the opposite direction as the Highlight wheel in my High Tint node. Usually I do the skin balancing after the main look for the grade has been set. But its always a push/pull back and forward balancing act to make sure everything in the grade gels.

Adding or retaining colour variation in the skin prevents the monotone plastic look that is all too common.

2. The other common thing I see is skin that has had the specular highlights graded out. This creates a flat lifeless look. Couple this with a monotone skin palette and your in a world of ugly.

A lot of the time making grading decisions that dont kill the specular highlights is all you need to do. But sometimes I find that creating a serial node off the main skin node that tints just the skin specular highlights is necessary.

I make sure that this node has the same exact qualification as the previous node. Then I set the High Range from between 0.500 – 0.666 and use the highlight colour wheel to tint the highs. Usually I add a bit of blue/purple and even teal. This plays off nicely with the yellow that was added to the highs in the main skin node. Sometimes I bump the brightness of the highlights a fraction to get them to pop more, but its not always necessary.

Skin Tone Examples

UPDATE: Someone on BMCuser wanted to see examples of plastic/bad skin tones and examples of good skin tones. Below are a few, it should be obvious which is which. All grabs were taken from the 1080p jpgs on Blu-ray.com except for the Hitchcock grab which was taken from the 1080p trailer on Apple Trailers.

These are not meant as direct comparisons, since there is different grading intentions, lighting etc.

And just to show you how bad it can get, I took one of the worst examples of monotone skin I could find and made one side completely monotone. Not much difference!

I’ll write up a proper post on grading skin with examples etc in the near future…

Click images for 1080p versions.

The post Blackmagic Cinema Camera + Davinci Resolve Colour Grading Breakdown appeared first on JuanMelara.com.au.

5D MKII + Davinci Resolve Colour Grading Demonstration from Juan Melara on Vimeo.

BMCuser.com member Ervani posted a request for members to try their hand at colouring his footage. Contemplating a switch to Davinci Resolve, Ervani wanted to see what Resolve users could do with his 5D MKII footage. You can read the original thread here.

The footage was shot with the Technicolor picture style, so it was ready to go inside Resolve with no extra work. If it had been shot with a regular profile, I would’ve used a video2log input LUT to convert the footage to log.

5D MKII footage starts to fall apart quite easily if pushed too far. I tried to be gentle with my corrections to avoid this, so the majority of the look is achieved without using secondary qualifications.

In the 1080p grabs below, there are all sorts of artifacts visible, which don’t seem so evident when in motion. 5D MKIIs are definitely not BMCC or Red cameras!

The video above includes a few before and afters. Click images for 1080p grabs.

The post 5D MKII + Davinci Resolve Colour Grading Demo appeared first on JuanMelara.com.au.

Kodak 2393 Emulation LUT Test + LUT Download Link from Juan Melara on Vimeo.

What is a LUT?

A look up table (LUT) mathematically translates a set values in your image into another set of values. For example converting a flat log image into video gamma that has correct brightness and contrast, or converting from one colour space into another. Another common use is to preview how your image will look when displayed on a target format or device.

Print film emulation

A film print emulation LUT previews how your image will look like when printed onto a target film stock. It accurately emulates the density and colour response of the film to give you an acurate preview on your grading monitor prior to going to print.

To achieve this accurately your monitor needs to be profiled so that the generated LUT takes into account your monitoring setup plus all the lab variables and the qualities of the final print stock.

This LUT is applied at the monitoring level so that your colour corrections / grading sit below it in the image display pipeline. Prior to going to print you remove the LUT and then output your project. If the LUT is accurate and has done its job, the resulting film print should be close to identical to the preview you’ve been viewing and grading on your monitor.

Print LUTs for aesthetic reasons and why you should use them

Even if your project is not going to print, you might still want to use a print LUT for the aesthetic effects it has on your footage.

Applying a print LUT is the probably the best “film look” treatment I’ve seen and the easiest way to make your digital footage look like film. What print LUTs do to the colours and highlight rolloff in your footage is about as close as I’ve seen to a one click “make awesome” button. Your images will still need work and you will still need to know how to colour correct and grade, but your work can look so much better.

Unlike a lot of products, presets, looks and tutorials I see, a print LUT is actually an accurate and measured emulation of film. And not someone’s interoperation of what they think film looks like. I’m often left wondering if the creators even know what film looks like, cause their presets, tutorials and samples most definitely do not look like it!

Print LUTs are easy to use, there are no variables or sliders to tweak and best of all they’re free!

Print LUTs for Resolve

I’ve had a lot of trouble finding print LUTs online. The LUTs I use, have usually come from a download link that has died years ago and I’ve been converting them from one format to another as I change grading packages.

So I guess you could say there is a real shortage of LUTs online, especially free ones. There are several paid LUTs that I’ve purchased and while they’re ok, I find they’re not as good as the LUTs you are about to get for free…

About a month ago Reduser member Bjorn Benckert from Syndicate Entertainment made available a set of print LUTs. Unfortunately they weren’t in the correct format for Resolve. So I took the time to convert them into a format Resolve reads. You can download these Resolve ready LUTs here:

Download the LUTs!

In the zip you get the following print film stock emulations:

• Fujifilm 3510
• Fujifilm 3513
• Kodak 2383
• Kodak 2393

Each LUT comes in three flavours: ConstLclip, ConstLmap and CUSPclip. Each of these treat out of gamut colours in a different way. For an in-depth description of each version, read page 9 of this document.

I mainly use the ConstLclip version of each LUT, but do your own tests. These LUTs use a generic HD profile as the target device they will be displayed on. So even though they don’t take into account your grading monitor since they haven’t been created especially for it, aslong as they’re viewed on a Rec709 calibrated monitor they should look pretty good. Actually they even look good on non calibrated PC monitors!

Installing the LUTs

On a Mac these LUTs need to be placed in Macintosh HD > Library > Application Support > Blackmagic Design > Davinci Resolve > LUT > CineSpace

On a PC place them in C:\ ProgramData \ Blackmagic Design \DaVinci Resolve \ Support \ LUT \ CineSpace

Setting up your footage in Resolve

The only requirement these LUTs have is that the source footage is in log gamma.

Red footage

For Red footage set the Gamma Curve to REDlog Film in the Resolve Camera Raw settings

BMCC footage

For BMCC footage set the Color Space and Gamma to BMD Film in the Resolve Camera Raw settings

5D and other Canon DSLRs

Shoot your footage in the Technicolor CineStyle profile.

Other cameras

If your camera doesn’t shoot log, or you have existing footage not in log, I have an input LUT that converts video gamma footage into log. If you want this LUT leave a comment or send me an email. I havent worked with it enough to be confident it works well with all footage, so I’m hesitant to release it. But if there’s enough interest I’ll test it thoroughly and put it up for download.

UPDATE: Here is the input LUT to convert video gamma into log. I haven’t had a chance to test it further so feedback welcome. Apply it to the footage by selecting it as the 3D Input Lookup Table. Or on the first node of the grade.

Applying the LUT

Once you’ve setup your footage as described above you’ll need to apply the LUT. Since you want the LUT baked into output footage, you’ll want to select the LUT as the 3D Output Lookup Table, like below:

Grading with the LUT applied

First thing you may notice after applying the LUT is that the footage looks contrasty and may be lacking saturation. This is normal. The first thing I do is make sure the image is correctly white balanced. Then with the colour wheels in log mode, I usually balance the image by raising the offset and lowering the highlights and increasing saturation. This is usually enough to get the image looking good. Sometimes using the Contrast and Pivot control plus saturation is all you need to do. From there grade as you normally would, but now enjoy film like colours!

UPDATE – real world examples

Interested in seeing what’s possible with these LUTs using real world production footage? Check out my new blog post here. Footage courtesy of Alex Montoya, see his Vimeo page here.

Comparisons

Below are a few samples. These are all grabs from Epic footage shot on the Samyang/Rokinon 85mm. The effects of the LUT should be the same on footage from any camera.

The first image of each scene is REDcolor3 / REDgamma3. I’ve white balanced and added contrast and saturation, basically just a first light – getting it ready for further grading.

For the second image I’ve left the white balance the same, switched to REDlog Film and applied the Kodak 2393 ConstLclip LUT and roughly matched the contrast to the previous image. By equalising the contrast you can get a better idea for what the LUT does to the colours.

The third image is a quick grade with a slight bit of film grain added. I’ve tried to keep the image looking neutral so don’t expect any stylised looks like you see on the rest of this blog!

It can be hard to see the differences between the images. I recommend opening each full size image in a new tab and flicking back and forth, the differences will become much more apparent. Pay particular attention to what the LUT does to the shadows, highlights, saturated colours and skintones.

Scene 1

Scene 2

Scene 3

A few more

When print LUTs may not be appropriate

Adding a print LUT actually degrades your image. You’ll notice it removes a lot of colour contrast and hue variation. And the film effect is actually quite strong. So if you’re after a clean, slick look or your after neutral images it might be easier to achieve without using the LUT.

Beyond print emulation LUTs

A print LUT is a one size fits all emulation that doesn’t take into account the colour response of the camera. A better solution for emulating film is FilmConvert from Rubber Monkey Software.

To quote the site “FilmConvert models the characteristics of the digital sensors, and provides a complex algorithm to transform digital footage to match your favorite film stocks – including negative densities, printer spectral responses and grain response curves.”

It works with Red cameras, Canon DSLRs and the GH2 plus it has a generic profile for use with non profiled cameras.

From testing the demo version it seems to match the film look a lot closer and faster than using the print LUT. Definitely give it a look.

FilmConvert – Digital vs Film comparison from Film Convert on Vimeo.

Problems?

If you’re having any problems working with the LUTs or can’t get them to look good with your footage, leave a comment or drop me an email and I’ll be happy to help you out.

The post Print film emulation LUTs for download appeared first on JuanMelara.com.au.

Kodak 2393 LUT Alex Montoya Footage Breakdowns from Juan Melara on Vimeo.

So how do the LUTs from my previous post look when used with footage from a serious production? Below are some real world examples using R3Ds from Alex Montoya’s great short film Maquillaje (Make Up). See the short film on Alex’s Vimeo page here.

If you’re interested in grading these R3Ds yourself, the original reduser.net thread with the R3Ds can be found here. These were all graded using the Kodak 2393 LUT.

As I always say, it’s really easy to grade well shot material. The quality of the source footage is a real testament to Alex, his cinematographer and everyone involved in the production.

About the grade

With this grade I was trying to reduce the palette to earthy browns and light greens, and at the same time giving more emphasise to any yellow in the scene.

I find it’s really difficult to judge how much saturation to add or keep in a grade. It’s very easy to fall in love with bright saturated colours as you grade, causing you to add more and more saturation as you go, to the detriment of how well the grade fits the content of the scene. I liken this to the anecdote of placing a frog in a pot of water on a stove and boiling the water. The water boils so slowly the frog never feels much of a change and never jumps out! It’s a good analogy and a good reminder to take breaks so you can review your work with fresh eyes.

I find its best to err on the side of less saturation. Looking at these grades now, I think it could probably be taken down another 5-10%.

You can see a breakdown of the grades in the video above. Click the images for full size grabs.

The post Kodak 2393 LUT – Alex Montoya Real World Examples appeared first on JuanMelara.com.au.

I’ve been busily working on the first tutorial in my video tutorial series. Here is a sneak peek of what I’ve been cooking up. I’ll be updating this page over the next 2 weeks with more sample images prior to the final launch, so check back often for the latest info.

The first tutorial in the series is my take on the summer blockbuster look. In the tutorial I break down the look, explain the colour theory behind why it works, I then breakdown the node structure behind the look in an easy to understand format and then show you exactly how to achieve it inside Resolve.

April 10 Update: Below is a Red Epic shot by Toby Linden that I grade in the tutorial. The video is taking a bit longer than expected, but I’m trying to put in the ground work early on, so I’ve got a strong foundation for future video tutorials.

Below is Red Epic 5K 2.4:1 footage of a watch I shot. In the tutorial I grade this footage to match the look of Mission Impossible – Ghost Protocol. I show you exactly how it’s done and will make available the original R3Ds so you can follow along.

Below the watch I’ve included some screengrabs of Ghost Protocol to show you the look I’m trying to match.

The post Summer Blockbuster Tutorial Sneak Peek appeared first on JuanMelara.com.au.

01 The Summer Blockbuster Colour Grading Tutorial from Juan Melara on Vimeo.

Introduction

I’ve received a lot of requests to share the techniques I use to achieve the various grades on this site and on Vimeo. But by far the biggest request has been to show how I create the cool/warm look. Otherwise known as the summer blockbuster look, the teal and orange look etc. The techniques I show you also form the foundation for most of the grades on this site. I hope to create a series of these tutorials (hence the 01 numbering) so stay tuned for more upcoming tutorials. I’ve already got a few ideas but If you have any suggestions, send me an email or contact me on twitter.

In this tutorial I assume you have some basic Resolve knowledge and you’re able to setup up projects, import footage, apply LUTs and create nodes. I also assume you have a basic grasp of colour correction and colour correction best practices. If you’re new to Resolve or colour correction/grading in general, I recommend you check out some of the resources at the bottom of this post.

Special thanks

Before I go any further I have to thank Toby Linden for kindly providing the footage of the gentleman with a gun. Toby is a talented cinematographer, check out more of his work at his website – dreamland-films.de and follow him on twitter – @TobiLinden.

Why Resolve?

Short answer: It’s really good, it’s free and it’s very easy to use.

The only excuses you should have for not using Resolve is if your system doesn’t have the specs to run it or you’re already using something better. In both cases you’ll still be able to get something out of the tutorial as the basic techniques should be applicable to whatever colour grading software you are using.

Why use film print emulation LUTs?

People ask me why I use film LUTs and if they are somehow cheating by using them. I use LUTs because I think they’re an important component in achieving this look. Film LUTs have been used by post production studios on films for aslong as they’ve been conducting DI colour grades. There’s a good chance that the films people continually try to match the look of, have been graded with a film LUT which has then been baked in, thus becoming a large part of the finished look.

Give the LUTs a try, see if you like their aesthetic. It should still be possible to achieve the looks from the tutorial without them, but it’ll be a little more difficult and I find the results generally aren’t as good.

So is it cheating, shouldn’t you be creating the look from scratch by hand? Well you are creating the look from scratch and by hand! The LUT is really only setting some rules for what colours are and aren’t possible, it’s not grading the image for you. When I grade for a client they only care about the end result. They don’t care how I get there or that I’m really printing out each frame in B&W and then colouring them in using crayons. As long as they get the look they want on time and on budget it doesn’t matter to them. The only people you impress by doing something from scratch and by hand are other colourists.

The tutorial

I tend to power through the tutorial, so you might need to re-watch certain parts. But I prefer to watch an information packed 5 minute tutorial 6 times, than a 30 minute tutorial that only has 5 minutes of worthwhile information.

I’m not sure, but I might be making it look easier than it is. Because of this I recommend trying the looks out on the same footage I use in the tutorial before trying it out on your own footage. This will allow you to get a feel for when you are heading in the right direction.

Ingredients

So if you want to follow along their are a few things you will need:

1. Davinci Resolve. Download the free Lite version from Blackmagic Design.
2. The footage I use. All the files in their original R3D format can be downloaded from here. If your system isn’t up to the task of handling 5k footage, let me know and I’ll look at creating 1080p/720p versions…
3. The film print emulation LUTs. Download them here and read this post for instructions on how to install and use them. If you are grading your own footage not in log format, download the video2log Input LUT from here.

The theory behind the 2 node subtractive colour setup

Analysing the blockbuster look its clear that it’s more complex than adding a cool hue to anything that isn’t skin tone or using the push/pull technique: cooling down the shadows, warming up the highlights and hope that your skintones look natural.

In the tutorial I analyse the look of Mission Impossible: Ghost Protocol and show you it’s not one single cool hue being added to the image, but actually several hues that appear at specific places on the luminosity ramp.

The subtractive colour model states that adding two complementary colours together cancels them both out and returns them to neutral. Using this knowledge, we place two log nodes in series which gives us 6 range adjustable control points on the luminosity ramp. We can then use those 6 control points to add hues exactly where we want them and cancel out hues we don’t want by adding the complementary colours. A powerful setup that is easy to tweak and adjust.

Questions, problems, feedback, complaints?

For any of the above leave a comment on this post, email me or get in touch via twitter – @JuanDMelara. If I get enough of the same question I’ll update this post with an FAQ.

Training resources

New to Davinci Resolve or colour grading? Take a look at the following high quality resources:

• Ripple Training – Davinci Resolve 9 Core Training
• Tao of Color - The ‘Grade-Along’ Training System for Resolve 9
• Color Grading Central – Davinci Resolve 9 Training
• fxphd - Davinci Resolve 9 Class
• Mixing Light - Color correction training. Monthly
• Lift Gamma Gain - The Colorist & Colorgrading Forum

The post The Summer Blockbuster Colour Grading Tutorial appeared first on JuanMelara.com.au.

What I really like about Davinci Resolve is the flexibility offered by the node based workflow. There are seemingly endless ways to setup the structure of your grade. And while I sometimes setup some complex node trees, they all generally follow the same basic plan. In this post I’ll show you this plan and explain why I use it for the majority of my grades.

Readers of this blog will already be familiar with this structure, but hopefully this post will shed more light onto ideas behind it.

So what’s the plan?

Above is the basic plan. Each box doesn’t necessarily represent a node but rather an operation that can be made up of any number of nodes. It also shows the order of operation from left to right (if that isn’t clear from the large numbers!?).

The basic idea behind this plan is to make corrections in a logical order, each preserving the required image data to properly make the next stage of corrections, and the corrections further on downstream.

To simplify this further into 3 stages. Think of stage one as a global balancing of the image, to lay a solid foundation for your grade. Stage 2 is localised corrections, both technical and aesthetic. Stage 3 is back to global corrections, sitting atop with the theory being that they tie everything together.

1. Balance

Every grade starts with balancing the image on the first node (or if your footage is raw – in the raw tab and the first node). Regardless of whether I’m after a rich saturated grade or a low contrast desaturated grade with milky blacks, I always balance the image with the same standard target in mind. To create a neutral, correctly exposed and suitably saturated image, full of information which makes downstream adjustments such as keying easier. A benefit of this standardised approach is the ease of copying over looks from other projects – a big time saver.

2. Localised Exposure

So once the image is balanced I move onto localised exposure corrections. Think of these corrections as an extension of the cinematography process but at the same time we’re able to push the corrections beyond that, and start to make creative exposure decisions.

Maybe the DP didn’t have ND grads to lower the exposure of the sky – I bring in a soft edged rectangular power window to take it down. Or maybe there wasn’t enough time to flag off one of the lights and there’s too much spill/bounce on what should be a dark background – an easy fix with a power window. The aim with these corrections is to be subtle and to make them look like they were achieved in-camera.

The reason I do these corrections at this early stage is because I quite often create looks that add colours to specific luminance values. If these values don’t exist prior to the application of the look then the colours from the look can’t be added where they need to be. Applying the exposure corrections after the look won’t yield the same natural looking results and can sometimes be a give away that an image has been manipulated.

Below is an example. Both of the results actually look somewhat ok, but it still illustrates my point.

1. The first image has been balanced and a look has been applied. The look is quite basic – purple added to the shadows, blue added to the mids and yellow/orange added to the highs.
2. In the second image a soft rectangular power window has been added after of the look to lower the exposure of the sky. Notice how it’s just taking the existing colours and lowering their luminance (and adding a bit of saturation in the process).
3. The third image is the exact same correction as in the second image, but this time it has been added before the look. Notice how the darker values are now picking up the colours from the look.

While both of these examples are neither right or wrong, I think it’s important to understand the difference the order of operations can make, allowing you to place your exposure corrections where they need to be for the desired result.

3. Localised Colour

This stage is similar to stage 2 but the corrections now focus on colour. It’s a combination of correcting elements that weren’t right at the time of shooting, such as correcting wardrobe colours etc. But it’s also creative, and is the first stage of building the look. Colours or elements such as skin tones are qualified, corrected, and then pushed where they need to be. This is usually a back and forth process carried out at the same time as building the bulk of the look in stage 4.

4. Look

By now the image should be both technically and aesthetically pleasing and in a lot of cases this fourth stage might not be necessary. But sometimes you’ll want to push the grade further. Think of this stage as an overall look you’re applying globally to the image (or most of the image). Such as warming the image by adding orange to the mids and highs or adding teal to the neutrals and cool colours to achieve the blockbuster look.

After adding the look you’ll find that sometimes some of the colours underneath might need to be adjusted. Maybe the skin is far too warm after adding the orange to the mids and highs. At this point I go back and make the required qualifications and adjustments in stage 3.

Like I said, it’s a back and forth process. Making mostly global changes in stage 4 and localised changes in stage 3 and even going back to stage 2 to make exposure adjustments.

5. Final Adjustments

In some ways this last stage is an extension of stage 4. This stage usually consists of making final adjustments to really glue all the stages together. Here is where I sometimes add a curve, make final exposure adjustments, add or remove contrast, lift and tint the blacks and tweak the overall saturation. This is usually carried out in a single node that sits right at the end of the grade. Think of this node as an adjustable LUT. It is applied to all the clips in the project and apart from saturation I usually don’t adjust it on a clip to clip basis. Keeping this last node mostly unchanged is an easy way to ensure there is consistency in the grade between all the clips.

Inside Resolve

Above is this plan in it’s most basic form as setup inside Resolve. This example is made up of only serial and parallel nodes. Of course you’re not limited to just these nodes.

If you’re not familiar with the different nodes available in Resolve, read the explanations below.

Serial Nodes

Serial nodes are easy to understand. Corrections are applied in series and added on top of the corrections from the previous node. In the example above I’ve balanced the image in node 1, this is then input into node 2. In node 2 I’ve used a circular power window to darken the edges of the image. Now the output of node 2 is the sum of the corrections from node 1 and node 2. The output of node 2 is now considered our master image, all corrections in stage 3 will be carried out on this master.

For the majority of adjustments serial nodes are all you need.

Parallel Nodes

Nodes setup in parallel apply their corrections to the input at the same time. This is also regardless of where they appear on the parallel stack. To illustrate the difference between serial and parallel nodes lets look at nodes 3 and 5 in the example above. In node 3 I’ve qualified the skin and pushed it a little warmer. In node 5 I’ve qualified the redder parts of the skin and pushed them further towards red. If I did this with serial nodes it would be difficult to qualify the reddest parts of the skin in node 5 since warming the skin in the previous node (node 3) might remove the the red from the skin entirely, making it difficult/impossible to qualify. This doesn’t occur in parallel since both nodes 3 and 5 have access to the same quality information being output from node 2.

Layer Nodes

Layer nodes have a similar appearance to parallel nodes, but their functions can be quite different. Think of layer nodes as a horizontal stack of corrections with priority given to nodes lower in the stack. To reveal the corrections of nodes higher in the stack, you need to qualify elements in the lower nodes. Anything outside of that qualification will show through. At the same time you can select the composite mode (add, multiply, overlay etc) that dictates how all the nodes are combined. I like to use layer nodes to split the image in 2, applying a set of corrections from multiple nodes to the image and then then bringing back uncorrected parts of the image and layering them on top. Check the Summer Blockbuster tutorial for an example of this.

For a better explanation of these nodes read page 468 of the Resolve 9 manual.

For a really good visual explanation of the differences between serial, parallel and layer nodes watch the following video by Gray Marshall and check out his blog.

Resolve 9: Parallel vs. Layer Nodes from Gray Marshall on Vimeo.

Order of operations

While its quite difficult to make destructive corrections which cant be reversed in downstream nodes, it’s still crucial to make corrections in an order that allows creative flexibility downstream. I’m quite often sent screenshots of grades where the user has applied a look too early in the node tree, sometimes even in the first node. This has then made it difficult to qualify elements such as skin in subsequent nodes. This issue can largely by avoided by making the corrections prior to the look (such as in parallel in stage 3). I also often see grades that didn’t balance the image adequately in the first node, leading to similar qualification issues. Fixing this issue is usually harder since making global changes to the image balance when a look has already been applied, can sometimes require the look to be rebuilt.

For this reason it’s important to treat the image with respect at each stage, to ensure you’re handing over good data to the following stage. This is a lot easier than trying to correct problems by adding more nodes downstream. After stage 4, its unlikely I will make colour based qualifications, so theres no issues blanketing the image with a global look and its precisely for this reason I leave the application of a look towards the end of the node tree.

Destructive corrections / clipping data

Speaking of destructive corrections there a few instances where Resolve allows you to destroy or clip data which can’t be recovered in subsequent nodes. Clipping occurs with the application of LUTs, Soft Clips and Vs Curves (regular Curves are fine). For this reason try and limit their usage towards the end of your grade where you won’t be needing the clipped data.

For a really good explanation of the instances that Resolve allows clipping, check out another video by Gray Marshall.

Resolve Tips: Accidental Clipping in DaVinci Resolve v9 from Gray Marshall on Vimeo.

This clipping of data is one of the reasons I recommend that LUTs be applied on the very last node or through the node tab, it just produces better results.

To illustrate this – below is an example of the same gamma correction applied after the LUT node (left) and before the LUT node (right). And whilst the LUT isn’t clipping the image into white, it’s still restricting the colours, which prevents them from getting richer as would be expected when lowering the gamma.

Conclusion

Whilst there really are no right or wrong ways to structure your grade, from experience I find there a structures that order corrections in a way that tend to produce consistently better results. If anything consider the order in which you make your corrections and question whether there is an easier, cleaner more logical way to achieve the same if not better result.

As always if you have any questions or comments use the comment form below or feel free to drop me an email.

The post Basic Resolve Node Structure and Order of Operations appeared first on JuanMelara.com.au.

Earlier this week John Brawley made available the first ProRes files from the Blackmagic Pocket Cinema Camera. Being a fan of it’s bigger brother, I had to take a closer look at them…

If you want to check them out for yourself, download the files from the link in this article on EOSHD.com. Theres also a few more available on John Brawley’s Copy.com account. The files in this post are G008_1_2013-08-06_0906_C0007.mov and G008_1_2013-07-27_1514_C0001.mov.

First up is a landscape scene. Below is the log ungraded image straight off the camera. Looks nice and flat, but a lot more desaturated than I would have expected/wanted…

Luckily there is a LOT of information to be dug out and pushed around. Below is a relatively natural grade from Resolve. It’s quite incredible the amount of information that lurks in the ProRes 422 HQ files! Who needs 444 and raw!?

Here is an outdoor scene with some skin tones. Below is the log ungraded image.

And above is another grade from Resolve.

In these two grabs I’m playing with a few different methods of adding saturation to the image. Ways that look a bit more organic and less heavy handed than just boosting the saturation value…

What I really like is the way the camera rolls the highlights into clipping. To my eye it looks very very filmic. I have made this a bit more gradual with a custom curve, but it looks pretty good straight out of the camera.

The full resolution crops do show a bit of moire, exaggerated by my abuse of post sharpening, but to my eye it’s not really any different to the moire in the BMCC footage. There is a little bit of noise present but it looks pretty good – most of it nice and tight rather than big and blotchy.

Like I mentioned before, I would prefer to see a bit more saturation in the log images. As this could possibly make better use of the ProRes codec.

Overall I’m impressed. It’s pretty incredible the quality of images coming off a sub US$1000 camera!

If there is any interest I might upload a grading breakdown video of both grades…

The post Blackmagic Pocket Cinema Camera – ProRes Resolve Grades appeared first on JuanMelara.com.au.

It’s been a while since my last blog post, I’ve been super busy shooting and grading various projects over the last few months. Luckily some of these are starting to see that light of day so I’m able to start sharing them.

Here are some stills from short film called Deep Water. It was directed by Daniel Zagaevsky with cinematography by Xiaolong Liu. You can check out Xiaolong’s other work here: http://www.xiaolongliu.com/

It was shot on the Alexa in LogC ProRes, which provides a great foundation for colour grading.

Unfortunately it hasn’t yet been released online. I’ll put up a link once its out.

Kodak 2393 LUT – D55 White Point

I used a Kodak 2393 LUT with a D55 (5500K) white point. Which results in a warmer image compared to the common D65 (6500k) white point. This was purely a creative choice – I liked what the warmer white point did to the skin tones, in particular in the interior scenes and also the way it shifts blues towards greens.

You’re not locked to this warm look for every shot since it can be quite convincingly graded neutral. Depending on how you do that, it can leave some tell tales signs of a warmer LUT, which I actually find quite pleasing.

My 2393 D55 LUT is custom, but if you want to give this a try yourself, Resolve 10 comes with various film emulation LUTs at D50, D55 and D65 white points. I actually really like these film emulations, they’re as good if not better than the ones provided on this site. Plus there are also P3 target versions.

Grading

The short is split into two locations each needing their own unique look. Theres the colder more bleak scenes that take place in Alaska and the more neutral scenes that take place in NYC.

For the Alaskan boat scenes I was inspired by the look of the below scene from Skyfall.

I ended up letting through more warmth from the LUT and pushing more green into mine (below) but I’ve tried to keep the same feel.

For the NYC interior scenes I let through the warmth from the practicals and enhanced them by adding a bit more warmth. For all the NYC scenes I used the technique I showed in the Summer Blockbuster tutorial to add coolness to neutral and cool colours. This time I only used the mid and shadow log wheels in one node, since the original footage already had a healthy amount of natural teal/orange so there was no need to push it.

The rest of the look was pretty straight forward, apart from experimenting with a few ways of adding saturation. I’m trying to stay away from simply bumping the saturation value which applies it to the entire scene. I’ve been playing with adding it non-linearly, adding more to the mids and less to the highs. Once I’ve got a something a bit more solid, I’ll share the the various techniques.

The post Deep Water Short Film appeared first on JuanMelara.com.au.

Based On True Events from Gary Allen on Vimeo.

Late last year I had the pleasure to grade Based On True Events a short film directed by Gary Allen, with cinematography by Zakaree Sandberg. It was shot on Red Epic 5k 16:9 for a 2.39:1 crop.

Finally here’s some proof that I don’t always apply heavy stylised looks!

The basic idea was to keep the look organic and natural looking. Removing the digital edge, but not going overboard with the heavy film emulation LUT look. I wanted the colours to appear vibrant where necessary and to shine through the grade.

I’ll update this post with some more grading information but for now here are some screenshots.

The post Based On True Events Short Film appeared first on JuanMelara.com.au.

In October I graded a great little short film titled Necrophilia: A Love Story. It was written and directed by Michael Millichamp. It’s currently doing the festival circuit, so it’s unlikely it will be available to view online for a good while. I am however allowed to share a few frames.

For more info on the short, check it’s Facebook page here.

The grade

Necrophilia: A Love Story was shot on the Red Scarlet in 4k.

Similar to Deep Water, I used a D55 white point film emulation LUT. But this time it was a custom Kodak 2383 LUT. The differences between the 2383 and 2393 LUTs are quite minor. The 2383 has slightly more lifted blacks and seems to impart a bit more of a look on the footage – mainly more blue/green in the shadows. Overall the 2393 seems to be a little more neutral in both colour and curve. I went with the 2383 LUT as I felt the 2393 was probably a little too clean and clear in the shadows, which I didn’t think suited this project.

As with pretty much all projects I work on, the raw footage looked great straight out of the camera. Therefore the idea behind the grade was to enhance what was originally captured. For the night time grave shots and interior party scenes I used the summer blockbuster technique. Mainly too boost the cool night time look, rather than to create a teal and orange colour separation.

Keeping it consistent

Using the blockbuster technique in some scenes and not others created the risk that the grades would not match. There are several ways to ensure there is consistency between scenes. Probably the easiest way of doing this is to make sure your shadows/blacks are all roughly the same tint and density. Although quite subtle, I introduced a bit of red/purple into the shadows of every shot. This created an anchor for every scene/grade. So no matter what was happening in the midtones and highs (within reason), the grades work as they appear to be built on the same foundation. It’s important to point out that the tint is only just visible in the area above black – the upper shadwows. Below a certain luminance as the shadows approach true black the tint is no longer really visible. I find its crucial to keep the deepest blacks tint free, as this retains a naturalness to the grade. Otherwise its easy to start heading into heavy tint instagram filter territory – which isn’t always desired.

The same idea applies to the highlights. Although I find it isn’t as crucial in achieving consistency as it is with the shadows.

How to…

So there are a few ways to introduce a tint into the upper shadows whilst still retaining relatively neutral blacks. The quickest and still relatively precise way of doing it is using the shadow log wheel. I usually do my shadow tints towards the end of the node graph usually in the last couple of nodes. On the first of the shadow tint nodes I add the main tint – purple in the example below (left). In another node downstream, I use the shadow log wheel again but this time with the low range adjusted to a value below the low range value in the previous node (right). Then I push the wheel towards the complementary colour of purple, which cancels out the purple tint but only in the lower range. Thus returning the blacks to neutral.

The other method I sometimes use is to setup the first tint node exactly the same, but on the next node I create a luminance qualifier to isolate just the blacks and lower shadows. Then I desaturate them. This method can work quite well although there are two things to be aware of:

• If the over all look of your grade is warm, desaturating your blacks will push them towards neutral. But neutral desaturated blacks will actually look cool in comparison to the rest of the warm grade – something which might not be desired.
• Desaturating your blacks returns them to neutral, but in doing so you also lose chroma (obviously). Again something that might not be desired. This isnt always obvious, but if your luminance qualification reaches too high up, it can be quite visible.

These are the benefits that the 2 node complementary colour setup has. 1. You can avoid having neutral blacks that don’t fit in with the overall colour temperature of your grade, as you can put them wherever you like – not true neutral but neutral within your grade’s colour palette. 2. And no matter how much of a tint you have in the first node, you can always return the lower shadows/blacks to neutral without losing any chroma.

The post Necrophilia: A Love Story – Short Film appeared first on JuanMelara.com.au.

Alexandros A. Apostolos on the Lift Gamma Gain forums posted this thread asking how to achieve the look of the Woody Allen films Midnight in Paris and To Rome with Love. Not seeing anything particularly complex in the look of either of the films, I decided to have quick go at trying to reverse engineer the grade.

Seeing as both looks are quite similar I picked Midnight in Paris as it was easier to track down images of the same locations. Alexandros was mainly interested in the outside shoots featured in the intro, so I’ll concentrate on those.

Getting Started

One of the first things I do when figuring out a look is to check IMDB to see if I can glean any info on the print stock used. It’s quite likely that the print stock emulation LUT used during the grading process is then baked into the non film release versions of the movie. Therefore playing a large part in the overall look of the Bluray, DVD etc versions.

IMDB states the print stock as Kodak 2383, so I’m going to use a Kodak 2383 LUT but with a warmer D55 whitepoint. With a bit of experience it’s quite easy to tell the whitepoint of the LUT used. The main tip off in Midnight in Paris is the overall warm cast throughout, particularly in the highlights – even in scenes that one would think would be neutral.

Resolve comes with several film look LUTs and actually has a D55 version of Kodak 2383. So using this LUT will put us on the right track…

The next step is to find suitable images to grade. It’s important to get as close to the original film images as possible. Looking at the intro sequence most shots are either overcast or rain soaked. Focusing on the overcast shots I tried to find images with matching location, angle and most importantly similar light conditions. I find Flickr is a great resource for this and a quick search yielded plenty of overcast images of Paris. Flickr is great because it allows you to download high res versions of the photos, so you can work in 1080p, UHD, 4k etc. Below each image I’ve included a link to the original photo on Flickr.

The Grade

Below is a screen capture from the film. The first thing I notice is how warm the shot is overall. But at the same time the foliage is quite vibrant and pushing towards cool green, cooler than you would expect when warming an image this much. The red from the awnings and the man’s polo are also extremely vibrant.

Looking at these images online – on this blog and perhaps next to other images you might have on your desktop, the warmth, the reds and greens undoubtably look too intense. But seen in the way they were intended I guarantee your eyes would become accustomed very quickly. And whilst they would still look warm, you would quickly buy into this warmer more vibrant world and it would start to look somewhat normal.

If it was me grading this I would be tempted to knock back the saturation on the whole image, especially in the greens and reds. But doing this I believe would remove some if not all of the uniqueness of the grade. And it would just look like regular old Paris.

Below is my version of the grade. The original photo on Flickr.

Here is the node structure:

Take note of where I placed my video2log conversion. I did this so late in the node graph so that I could do the majority of the work in the image’s original SRGB gamma. When dealing with 8bit images sometimes I find a video2log conversion too early on in the graph can introduce artifacts as you’re pushing the image’s 8bits too hard and then making corrections on top of that.

Also take note of how the foliage is split into two nodes. One node cools most of the foliage but especially the darks, the other node pushes the highlights warmer and more vibrant. Something you wont be able to see is the warmth in the image is being added via the offset control in the Warm node.

And here is a video breakdown showing the effect of each node:

Midnight in Paris Grade Breakdown from Juan Melara on Vimeo.

Below are a few more screen captures and my matched grades. I think a measure of how well you’ve reverse engineered a grade is how easily it can be applied to other source material. Very little tweaks were needed for all these images beyond the standard balancing in node 1.

Above – screen capture from the film.

Above – matched grade. Original photo on Flickr.

Above – screen capture from the film.

Above – matched grade. Original photo on Flickr.

Above – grade applied to another source image. Original photo on Flickr.

Above – grade applied to another source image. Original photo on Flickr.

The post Midnight in Paris – Reverse Engineering the Grade appeared first on JuanMelara.com.au.