Why grading Phantom footage sucks in Resolve

Grading Phantom footage in Resolve sucks - and there's a specific reason why. In this video I'll show you where the problems come from and show you how to use a free tool my community and I developed to make grading Phantom footage easy - and beautiful.
Grab the free Phantom IDT here: procolor.ist/free-phantom-idt
--------
Grab my Voyager LUT Pack. 17 LUTs to provide beautiful looks for any type of project so you can grade faster, take every image further, and attract better and better jobs.
procolor.ist/voyager-lut-pack
Get my free Kodak 2383 film print LUT for DWG and ACES here:
procolor.ist/freelut
Check out my ebook, The Colorist's 10 Commandments:
procolor.ist/ebook
Grab $100 off the Raw Film Library
raw.film/lifetime?code=cullen...

[@thatcherfreeman]

For everyone's info, basically what's wrong with the usual Phantom color management is that Resolve decodes the Phantom raw clips to rec709 via the Phantom RAW API, and then Resolve just does a CST from rec709 to your timeline color space. The issue is that in the rec709 decode, their RAW decoder applies a creative or preferential tone mapping adjustment that's specified by one of the metadata flags in the RAW clips, and Resolve's CST does not invert this curve so it's baked in and the contrast is messed up. When you decode to Phantom Log 1/2, this tone mapping step is skipped, but guess what? Neither of those log curves are supported in Resolve because Phantom hasn't released the specs for them. So the work here was essentially just profiling the two log curves and converting to DWG/DI based on the reference ACES decode provided by the Phantom API.

[@MaximoJoshua]

Awesome work Thatcher! Are you going to have a video about the process you used? Could you do the same thing for DLog-M? Also, was that what was causing the highlight issue? They clip it down to Rec709 (losing the detail) then expand it back into Davinci Intermediate, it sounded like there were two things wrong with the built in method, the tone mapping being baked in, but also taking footage to 709 and then back out to DI...

[@thatcherfreeman]

@@MaximoJoshua Seems like sometimes my comments get deleted if I mention certain words. I'll eventually make a video about camera profiling, that's on my todo list, but i haven't done it yet. I've also profiled a few different DJI cameras, it turns out their D-Log and D-Log M curves differ for each model of camera, so you have to build a new IDT every time they release a new camera, which sucks.

[@thatcherfreeman]

@@MaximoJoshua The issue with the highlights was that the tone curve Phantom applies for Rec709 decode has a contrast boost and also a highlight rolloff baked into it. You can actually check out the curve in the PCC software if you have any Phanton clips, it's not particularly smooth, nor is it something you'd want in an IDT.

[@MaximoJoshua]

@@thatcherfreeman Yeah, i've noticed the same thing with comments, so annoying. DJI is so annoying with DLog-M, at least they include HLG, which is published and known. The weird thing is that they provide the identical LUT for every camera that has DLog-M. I compared them all, they don't even change the comments inside the LUT...they all say # Mavic 3 Pro, D-Log M, 2023-03-24 only the file name changes. This is the same on the Air 3, Osmo Action 4, Pocket 3 and Mini 4 Pro...

[@MaximoJoshua]

@@thatcherfreeman Oh, also if you need any Mini 4 Pro footage, let me know, happy to help.

[@christianrushpostpro]

the fact that this came just a few days before a huge commercial gig that is using a ton of Phantom footage. thank you so much Thatcher, Cullen, and anyone else that contributed to this.

[@luciadcolor]

Awesome Cullen! So great you came out with this solution. Thank you Thatcher! And thanks for sharing!!

[@andycrystal]

Great stuff as usual and also thank you @thatcherfreeman for explaining the under the hood issues of why this happens!

[@samkeythecomedian8616]

This is really helpful….. Thanks for sharing your knowledge here

[@garethpearce5255]

Ohhhh, well this is super handy. I'm going to be working on some phantom footage in a couple of weeks👍 Thanks guys!

[@cansenuk]

The title of this video had me thinking I was going to sit through 18 mins of the problem, thankfully you had the solution in there too! 😂

[@jhonesreece4946]

Very good tutorial on the color grading in DaVinci résolve 🎉🎉🎉 🎉🥳

[@lemaproduction8429]

Thank you guys for this. We've been working with a VEO4K since it has came up, and have always struggled dealing with the highlights and felt this weird shift. The two things which helped us were using a lot of soft lights to fight against this aggressive curve, and at the same time lighting the camera for 1600 ISO in order to get the best highlight roll-off. Not being an expert at color grading, I have two questions : how can I use this workflow while we are also color grading Red footage ? How could we export a LUT to better visualize the phantom footage on the set ? If you need some images to help you in going more in depth in your phantom workflow, let me know I will be glad to provide some.

[@Owlbot]

Yo brotha! I can hear you're using the Adobe Podcast Enhance situation, I do the same on my podcast as it's a miracle worker and many of my guests are just going full laptop mic no headphones "yolo" style, but I think if you get your mic a little closer to ya (maybe do a split comp if you need it in the frame?) it'll sound a lot more natural! Just bringing it up cuz I've been juggling it for a couple years haha

[@lfwesanderson]

Hey Cullen - I thought it might be a good shoutout to do a similar workflow for Flog2, a very popular Gamma curve from the Fuji cameras that's still not supported by Davinci even 2 years after its release. Fuji have published their white paper on Flog2 so I'm positive the data for the IDT is out there.

[@thatcherfreeman]

Ya we got that one in the bag, I'll ask him to release that one next

[@nuoxu2922]

Thanks for the great video! By the way, is the Phantom Cine footage you used available for purchase or download?

[@viggosundvold]

Hi Cullen, thanks for another helpful video! This reminds me of working with the new GoPro 12. This camera now has a new log format (10bit GP-log), that is not supported in Resolve CST. Would you consider making a similar video about working with this camera in a colour managed workflow? GoPros are so often used together with much more professional cameras, and is always horrible to match. I am hopeful about the new GoPro 12, but I can not input map it!

[@sashazephiria3869]

Hi Cullen ! So awesome you came up with a custom solution for that ! I've recently had to work on a project shot on Kinefinity, same issue, no input transform available. I didn't want to get stuck with the manufacturer's 709 LUT (which would make me unable to use DWG and your creative LUTs without treating the footage as 709 by doing : 709 LUT > CST 709 to DWG > Grade > DWG to 709). So I ended up using Arri's color science since it gave pretty similar results to the 709 LUT. But... still not the ideal method. Would love to get a custom input for Kinefinity (and other cameras that are not natively supported by Resolve :D )

[@thatcherfreeman]

I'm still working on that one, Kinefinity owners seem to take their time with getting footage back to me ngl

[@he.smile_]

@@thatcherfreeman the awesome folks at @team2films own a Kinefinity! Maybe they could help?

[@MaximoJoshua]

@@he.smile_ I was thinking the same thing, they made a video on the Mavo...they are an awesome resolve resource, so they might have a ton of footage already

[@filmowaniezprofitem]

I have similar problem with D-Log from DJI drones.

[@amswil3232]

Hi cullen i love your video and i have been watching them since day one, i just have one question , my brother's friend got the dehancer plugin for davinci and i have been trying to figure out how to grade in aces using your sandwitch aces method, i would be really grateful if i get an answer

[@arqrenderz6210]

Hi Cullen, I'm new to the channel but its a hell of a ride!! Do you have some tips for CG footage? I have looked but there is no reliable source of color grading science for cg shoots :(

[@brown2840]

Outside of certain specific circumstances like this (most of us prob aren't grading phantom footage for example) I usually just let resolve color manage in DWG. That way it interprets all of my braw footage from the get go and no CSTs are needed. If there are still shots that aren't raw, can't they still be tagged or converted properly with CST or no? I guess what I'm asking is why not let resolve handle as much as it can automatically upfront, then handle the rest manually? Is it mostly about visually knowing what you're controlling in nodes or is there something else I'm missing? Hope that makes some sense!

[@joshuagonzalez7819]

I really enjoyed this video. Excited to start working with this workflow. Just quick question. Do you guys know if the webpage to get the IDT is working? Im not being able to get there.

[@CullenKelly]

Glad to hear it Joshua! And we'll have the webpage with the IDT back up ASAP!

[@thomlyons]

you mention towards the end of the video that you would leave a link to both the 2383 and the 3510 FPE but I only see the 2383 one. Is there another link somewhere we could get?

[@CullenKelly]

Great question! If you look in the description of the video there is a link!

[@Visethelegend]

But if the timeline is already outputting to Rec 709 G2.2 why doing the extra step of having that Cst node at timeline level. Shouldn’t they cancel both out? What am I missing?

[@thatcherfreeman]

When using YRGB unmanaged, the output color space setting in the project settings basically just specified what NCLC tags are written to the file at the delivery page and doesn't actually change any of the code values that are written. So you still have to place a CST to the output color space at the end of your pipeline.

[@RasMarty_]

CM in nodes is the only way

[@IsaacRC]

I love this, i wish I could profile all my cameras with bad color science and no, color checkers doesn't magically fix bad color science by itself

[@thatcherfreeman]

My profiling process isn't much more than an exposure bracket to profile the log curve and a colorchecker to estimate the "primaries". It's true though you can't profile a camera from just one shot of a colorchecker though.

[@IsaacRC]

@@thatcherfreeman a bracket exposure to profile a log, i saw this before, if i may ask, why multiple exposure? Does a log curve vary?

[@thatcherfreeman]

@@IsaacRC Yeah so if you look at the Arri Log C3 white paper (or any of the other log white papers), you'll see that there are like 7 or 8 different parameters that you have to fit and log curves are usually a piecewise combination of two functions, with a scaled log in the highlights and a more linear portion in the shadows. Ultimately, you want to get to a point where you can convert from log to linear and do one-stop exposure adjustments in increments of 2x gain. The exposure bracket is to make sure the fitted log curve is correct across the dynamic range of the camera, though technically that's just for redundancy. You could get away with just two exposures that extend into the shadows and highlights, or with one exposure of the xyla 21 test chart, but it's cheaper and probably more robust to shoot an ~8-stop bracket and make sure you can correct the exposures. Hope that answered your question!

[@IsaacRC]

@@thatcherfreeman Far out! I ignore most of the math, i was profiling log curves by hand from 10 or more 1 stop brackets and comparing them with davinci's raw basic gammas like linear or sRGB. I see there's quite some math and complexity involved 📚