Powergrade for Resolve

[togg]

Sooo!

You export the dngs from MLV APP. Then what?

By following a couple of juan melara tutorial on youtubes and using is film lut I changed my post production process again!

No more 3d export lut in the preferences. I've moved into the color transform nodes.

I made 3 powergrades at the moment, the general idea is to put your wb in the raw panel, bmd film, +2 exposure (not needed if you expose correctly MAYBE) then first node a color transform to bmd to alexa log c just because it's more cool, then your corrections node, then another color transform where you bring it into rec709, apply the tonemapping options there, and then if you want add SOME of the nodes (not the contrast curve one) of the kodak poower grade that melare recreated.

One of the 3 is only the color transform and some clean contrast, another the color transform and some lifted contrast, a third the color transform and the extra kodak nodes that make film saturation and stuff.

Try them out!

https://www.mediafire.com/file/ojcqpc76pax6brz/MLV_dngs_color_transform_power_grades_for_davinci_resolve.zip/file

[Dmytro_ua]

Could you share a preview image of the RAW file and your edit result?

[togg]

Could you share a preview image of the RAW file and your edit result?

yes of course! What's your usual workflow with raw images? It would make a better comparison.

A basic Rec709 from the raw panel.




The 3 powergrades I did.

[cmh]

The grade 1.50.1 (third still I presume) is pretty simple but it has really great skin tones on an EOS M (the reds are leaning towards the orange tho).

My previous workflow for a quick first pass grade:
I used to enable Pre Tone Curve in the raw project settings, leave the color space to rec.709/sRGB and up the gain to +20 in the raw tab with few other settings: Saturation -5, Color boost +10 to temper the red channel a bit, Highlight -30, Shadows +10. Finally a gamut mapping node: Gamma sRGB, Tone Mapping Luminance, Max Input 1000.
It doesn't look bad but the problem is that if I do a daylight white balance of a mixed lights environnement, skin tones might have to be fixed separately.

Your grade is pretty spot on right off the bat. Thanks for sharing. It's a nice improvement compared to what I used to do, I'll try to fiddle a bit with it.

[cmh]

Testing this powergrade real quick in a mixed light scene (It's an EOS M, 1488x1866 squeezed, 12 bit).
I denoised, changed the white balance, upped the gamma a notch and that's it:
Not a valid youtube URL

[togg]

The grade 1.50.1 (third still I presume) is pretty simple but it has really great skin tones on an EOS M (the reds are leaning towards the orange tho).

My previous workflow for a quick first pass grade:
I used to enable Pre Tone Curve in the raw project settings, leave the color space to rec.709/sRGB and up the gain to +20 in the raw tab with few other settings: Saturation -5, Color boost +10 to temper the red channel a bit, Highlight -30, Shadows +10. Finally a gamut mapping node: Gamma sRGB, Tone Mapping Luminance, Max Input 1000.
It doesn't look bad but the problem is that if I do a daylight white balance of a mixed lights environnement, skin tones might have to be fixed separately.

Your grade is pretty spot on right off the bat. Thanks for sharing. It's a nice improvement compared to what I used to do, I'll try to fiddle a bit with it.

That is so nice to hear! The video looks great! I'm happy to have contributed some to the color science of ML raw Even if only by putting together stuff from other sources.

To recap what this powergrades basically do : 

- Use the transform LUT available into Resolve to map luminance and saturation in a more filmic way.

- Brings exposure up by 2 stops in the raw panel to basically increase the dinamic range and match what you see inside MLV App.

- Adds some contrast on a middle node (plus a color shift).


All of this can be vastly improved imho. There're tons of ways the tone mapping could be better probably but that's serious stuff. The 2 stops hack bring out some excessive noise, but how to do it differently ?

It would be nice to have a thread with better organized collection of power grades for Resolve. On the end that's what defines the image and it can be pretty discomforting for a new user to have to do this from scratch.

[cmh]

That is so nice to hear! The video looks great! I'm happy to have contributed some to the color science of ML raw Even if only by putting together stuff from other sources.

Yeah I'm in no way a colorist but I just wanted to say that despit not beeing "neutral" your powergrades are pretty natural and photographic, there's usually not much feedbacks on the ML forum and I wanted to show my appreciation.

All of this can be vastly improved imho. There're tons of ways the tone mapping could be better probably but that's serious stuff. The 2 stops hack bring out some excessive noise, but how to do it differently ?

I don't think that restoring the two stops in the raw tab increases noise as it just compensate for the bmd gamma which brings everything all the way to the left.
As far as I don' try to recover information like upping the shadows, it's ok (but then I'm using an EOS M that is known to have to pretty bad dynamic range so maybe I'm wrong).

I previously tried multiple tonemapping workflows, a decent way was to linearize everything and use Corona highlight compression with Tonemapping Tools in Fusion (filmic and Reinhardt are also available) but that was pretty slow and required me to up the gain instead of ISO otherwise I ended up with muted colors.
Using a colorspace transform node to go from bmd to rec.709 is way cleaner anyway and the gamma mapping is customizable enough for me.

It would be nice to have a thread with better organized collection of power grades for Resolve.

I'd love to see such thread, I would participate.

On the end that's what defines the image and it can be pretty discomforting for a new user to have to do this from scratch.

As for empowering newcomers on this topic, as cynical as it sounds, the large majority wants the youtube "cinematic" aka slap a LUT like an instagram filter. There's a steep learning curve, it would have to be prefaced by how to read the scopes instead of eyeballing it on uncalibrated monitors, yadi yada.

[cmh]

Just a basic color correction on random scenes using the power grade you shared to see how it holds up with skin tones in different conditions (an olive complexion shot at 5000K at multiple times of the day).
I'm using the 1.50 dpx and the Rec709 Fujifilm 3513DI D65 lut at 30% as a last node instead of Juan Melara's lut.
Fujifilm 3513DI is way too green in the shadows. I'll stick to Kodak 2383 next time and correct the red cast I get from it.
I tried to get rid of the Alexa CST multiple times but I've always ended up correcting highlights here and there using curves and tone mapping so I'll just keep it as it is.

Not a valid youtube URL

[togg]

Just a basic color correction on random scenes using the power grade you shared to see how it holds up with skin tones in different conditions (an olive complexion shot at 5000K at multiple times of the day).
I'm using the 1.50 dpx and the Rec709 Fujifilm 3513DI D65 lut at 30% as a last node instead of Juan Melara's lut.
Fujifilm 3513DI is way too green in the shadows. I'll stick to Kodak 2383 next time and correct the red cast I get from it.
I tried to get rid of the Alexa CST multiple times but I've always ended up correcting highlights here and there using curves and tone mapping so I'll just keep it as it is.

Nice! Again, it looks very good.
Also remember my middle node is pushing green up a little bit. I merged two contradictory saturation modifications together and added your final lut suggestion at 20%: https://www.mediafire.com/view/5n19fdwqzz12bhm/Powergrade_Resolve_1.7.1.dpx/file

It looks good on my end as well! But I should do more test. Also it appear clear that you can have the same looking image with the same powergrafe but with less noise by exposing it +2 stops in camera, and remove the raw panel +2. You'll just loose the highlights... Our choice.

[cmh]

I stumbled on some info on cdng and bmdfilm: https://forum.blackmagicdesign.com/viewtopic.php?f=2&t=87576

CaptainHook at Blackmagic Design
"Blackmagic Film" for Gen 1 is actually not a transform but passes out sensor space, or sensor RGB. You can basically think of it as "no colour science applied". So you would need a transform from that particular sensors response to the gamut of your choice which you'd need to get from that manufacturer. I believe Digital Bolex for example offered LUTs to transform from their sensor RGB to common gamuts like 709 via a LUT to be used in this workflow with DNGs in Resolve."
"Yes, DNGs in the metadata provide the matrices and AsShotNeutral tags needed to convert from XYZ to SensorRGB (and back) as supplied by the camera manufacturer so Resolve will be converting from SensorRGB -> XYZ -> Rec709. As for getting an XYZ output option I'm not part of the Resolve team so you would have better luck requesting that direct to them, but you could use CST to go from Rec709 primaries to any other gamut as the output from the RAW decode should not be clipped."

Jacob Fenn
"My issue ended up most likely being that the DNG's weren't relaying the ISO data needed for Resolve to properly debayer to Blackmagic's log curve. Therefore, when I'd apply what I thought was a normalizing transform via a CST node, the result was off because that CST node expects ISO 800. Thanks CaptainHook for help figuring that out."

To my understanding:

• There is no information clipped if we use rec709 (or linear whatever) in the raw tab and it's totally okay to, before any grade, use the first node as a color space tranform to log (like rec709 to alexa or cineon).
• Using bmdfilm in the raw tab is a no-no and needs further color processing (a lut to match this particular canon camera for exemple like Digital Bolex do) but to verify my statement we should both shoot a color chart (which I don't have) in the same lighting conditions and compare the results of our cameras with your 1.50 dpx for exemple.
• There's probably missing metadata in our dng compared to blackmagic cameras cdng, like ISO apparently that would explain the fact that we have to change exposure (as we said before the whole histogram is on the left which is not what a blackmagic camera would get).

edit: It might be the value named BaselineExposure. Compared with exiftool it is set to 3.09 on an ursa mini, 0.76 on a Digital Bolex D16 and 0 on an eos m cdng from MLVApp.
Tf you want to overwrite the original files with a value of 1.25 so you don't have to correct for exposure:
exiftool -BaselineExposure=1.25 -overwrite_original *.dng
It is a quick operation.

[togg]

To my understanding:

• There is no information clipped if we use rec709 (or linear whatever) in the raw tab and it's totally okay to, before any grade, use the first node as a color space tranform to log (like rec709 to alexa or cineon).
• Using bmdfilm in the raw tab is a no-no and needs further color processing (a lut to match this particular canon camera for exemple like Digital Bolex do) but to verify my statement we should both shoot a color chart (which I don't have) in the same lighting conditions and compare the results of our cameras with your 1.50 dpx for exemple.
• There's probably missing metadata in our dng compared to blackmagic cameras cdng, like ISO apparently that would explain the fact that we have to change exposure (as we said before the whole histogram is on the left which is not what a blackmagic camera would get).

edit: It might be the value named BaselineExposure. Compared with exiftool it is set to 3.09 on an ursa mini, 0.76 on a Digital Bolex D16 and 0 on an eos m cdng from MLVApp.
Tf you want to overwrite the original files with a value of 1.25 so you don't have to correct for exposure:
exiftool -BaselineExposure=1.25 -overwrite_original *.dng
It is a quick operation.

Daamn! Now I have to try all of this... You are really getting to the bottom of this exposure thing.

In the meantime I was helped to replicate some powergrade to replicate halation, if someone is interested it's here (setted for bmd film): https://mega.nz/file/x6xiybDC#7tpevoD-eSs6AE1sV4aHhVYLKN4RLM-6298kK5GQQXQ

[cmh]

Thanks I'll check that later in the day.

[cmh]

Sorry, I've been busy, I love the effect, really reminiscent of film. I find your powergrade less in your face than many or the tutorial available on the internet. Really cool.

[byrd]

Hi guys, thank you for this fascinating and helpful discussion! I have also been impressed by the grading tutorials from Juan Melara as well as yourselves, and I wanted to ask a related workflow question: Has anyone compared the results between this method starting with DNGs and doing everything in Resolve with exporting ProRes from MLVapp using the Alexa Log C profile?

I'm totally convinced by grading in Arri Log C, but the reason I ask is because I was reading over on the MLV RAW on EOS M facebook group that MLVapp has much better debayering and upscaling capabilities than Resolve. So I've done a few comparisons and to my not very well trained eye, it seems like exporting ProRes 422 HQ in a log profile from MLVapp does yield slightly better results, especially if you're upscaling a lot. And of course it makes for smoother playback in Resolve than working with CinemaDNG, which is nice if you're editing in Resolve too before grading.

Curious to hear your thoughts!

[wib]

Hi. it depends in wish resolution you shot. in my experience i had the bad surprise to notice some heavy aliasing in dng, but when rendered as mov from mlvapp, all the small details were smooth as butter.

So I think it’s maybe a better workflow this way : alexa log profile in mlvapp and juan’s lut in your favorite editing/color software.

[byrd]

Thanks wib! I'm often shooting in 1080 mode on the EOS M, so aliasing is also a danger for me and I kind of found the same as you describe, probably because of the better debayering and upscaling in MLVApp? I'm also just using the free version of Resolve, so the superscaling function isn't available.

However, I asked this question in the MLV RAW on EOS M facebook group yesterday too and Markus Schweitzer, one of the developers of MLVApp, pointed out that this practice will reduce the bit depth to 10bit, so that by doing so I'm actually losing a lot of information, and that LOG especially needs high bitdepth for the midtones. So he said it could result in banding. 

I guess I need to do more tests to see how the advantages and disadvantages play out in my own work. Markus's point was that it's better to do all the RAW processing and color correction in the same place, so either Resolve or MLVApp.

[allemyr]

I'm totally convinced by grading in Arri Log C, but the reason I ask is because I was reading over on the MLV RAW on EOS M facebook group that MLVapp has much better debayering and upscaling capabilities than Resolve. So I've done a few comparisons and to my not very well trained eye, it seems like exporting ProRes 422 HQ in a log profile from MLVapp does yield slightly better results, especially if you're upscaling a lot. And of course it makes for smoother playback in Resolve than working with CinemaDNG, which is nice if you're editing in Resolve too before grading.

Curious to hear your thoughts!

I think you can have the same quality in Resolve, but when using Resolve for the first time it has so many settings that's not set right so you have to do that first, I think that's wheres the problem lays. Its much fiddling before you can export your first movie. But when you have learned Resolve and setting it up you are done and can start us it. But yeah MLVapp is set up right from the start so its much faster!

[togg]

We should try to test shooting a chart as he does to see how the presets we are using are working with ML and where the discrepancies lies:

[wib]

Thanks wib! I'm often shooting in 1080 mode (...) that this practice will reduce the bit depth to 10bit (...) it could result in banding.

Yes, I forgot to mention that the strong aliasing in anamorphic was in 1080, in my 5D3. But not as much in 5K on thin and diagonal lines. Also, after a few test, yes unfortunatly I had some severe banding, even with prores 4444. So I guess from now, I'll stick to 3.5K and DNG, as the color are more richer and allows more control over the highlights in davinci.

[allemyr]

Yes, I forgot to mention that the strong aliasing in anamorphic was in 1080, in my 5D3. But not as much in 5K on thin and diagonal lines. Also, after a few test, yes unfortunatly I had some severe banding, even with prores 4444. So I guess from now, I'll stick to 3.5K and DNG, as the color are more richer and allows more control over the highlights in davinci.

If you link a video its easier to see and follow

[wib]

here for example, mostly visible at 30s, the guy walking with a grey backpack, his bag curve is like a staircase. It's more noticeable if you watch it on a 4K monitor

(shot in 1080p but anamorphic preset)

[allemyr]

Yes and at the shot with the bus at 23 sec its even more noticable. Other then that it looks really good I like the grading and I think I really like the quality of the lenses you are using, it is some old ones? Now I saw you vimeo comment. Good lens quality.

What resolution height and width are you shooting at and what do you upscale to? UHD 3840x2160 in the end?

[allemyr]

I post this here hope thats ok. Is there anyway how you can share a Resolve project with settings and all have find any solution yet.

Just wanted to share since I enjoy editing raw video files and the power of Resolve the more you learn it. In past years I would have had problems with a shot like this directly at the sun at 400mm with bad edges and so on. Happy as always with grading and my wallet is happy that I'am not looking at another more expensive camera then my 8 years old 5D3

Resolve 3840 -> 960 width


Resolve center crop 1:1


Adobe Camera RAW default


Resolve Hue vs Hue Curve


Resolve Hue vs Saturation Curve


Resolve Hue vs Luminance Curve


Will post a video soon, happy to see so much highquality grades lately on the forum and many by shots from 5D3.

[wib]

Yes and at the shot with the bus at 23 sec its even more noticable. Other then that it looks really good I like the grading and I think I really like the quality of the lenses you are using, it is some old ones? Now I saw you vimeo comment. Good lens quality.

What resolution height and width are you shooting at and what do you upscale to? UHD 3840x2160 in the end?

Thank you so much !

I can't answer your question, sadly I didn't kept the files, as it was for test purposes. But I'm certain that to get full POV I must have shot in anamorphic 5K. Or in mv1080. I think it must be 1080 as I noticed the strong aliasing in Davinci's debayering the DNG compared to MLVapp, and I didn't received yet a faster SD card to use card spanning.

Anyway yes, after I upscaled to UHD for upload on Vimeo.

But for the other movie "Paris libéré #2" I shot directly in 5K and 3.5K crop, and I didn't had aliasing pb's. Now that I am writing that, I'm pretty sure it was mv1080 for the "bike streets/champs" movie.

Here is the old lens and the mount I bough on amazon :

[togg]

here for example, mostly visible at 30s, the guy walking with a grey backpack, his bag curve is like a staircase. It's more noticeable if you watch it on a 4K monitor

(shot in 1080p but anamorphic preset)

so are you using some kind of variation of our presets? or going from alexa export from mlv app then Juan on resolve?