Anyone struggling colour grading n-raw n-log footage with LUTs

[Andre Yew]

Same sensor, same raw dynamic range. You only have to change the ISO strategy to match it. It'll only be different if they switched to a 14-bit readout for video.

Nikon rep answered this question in the ProAV vid at 1:05:30 -- DR is the same as Z6iii when shooting in Nikon mode, but NR is higher only when using R3D-NE -- RED WideGamutRGB/Log3G10 codex. In all cases RED-NE is 12-bit not 16-bit.

Raw is raw.

One possibility is that the CFA (color filter array) response the Resolve decoder is expecting from an R3D NE file and an N-raw file could be different. That is, the ZR uses a different CFA than the Z6III, but that'd be a thing to test when the camera is out.

Some people have reported different color renderings between the R3D and N-raw rendering pipelines when doing the renaming hack, and Leeming's even released a LUT that corrects for this.

 

[Andre Yew]

I do not fully understand the difference between the 33 (Monitoring) LUT and the 65 - Gerald clearly recommends use of the 65 or 65LC LUT when color grading -- they certainly do a good job.

Monitoring LUTs are made smaller so they can be applied in realtime by modest hardware, like your average on-camera monitor or an internal camera renderer (like the Panasonics have). It depends on the monitor: I've seen 65-point LUTs bring a Hollyland monitor to its knees, while it could handle the 33-point LUT fine for realtime monitoring.

--
https://www.instagram.com/lolcar/

 

[jlafferty]

I just played around with Davinci again looking at the same clips in .NEV and .R3D, with the aim of using the Nikon+RED LUTs, versus .R3D, and the color management settings for either are wildly different.

I could not get the .R3D renamed files to play with the Nikon LUTs nicely, no matter what hoops I jumped through with my color management settings.

If you go to RED's website and download their LUTs, you get some LUT names that are very similar to the Nikon package:

RED_FilmBiasOffset_to_Rec709_BT1886.cube

RED_FilmBiasBleachBypass_to_Rec709_BT1886.cube

RED_FIlmBias_to_Rec709_BT1886.cube

RED_Achromic_to_Rec709_BT1886.cube

You get a bunch others as well. These, in comparison, are the first few Nikon+RED LUTs:

RED_FilmBiasOffset_Rec2020_N-Log_to_Rec709_BT1886.cube

RED_FilmBiasBleachBypass_Rec2020_N-Log_to_Rec709_BT1886.cube

RED_FilmBias_Rec2020_N-Log_to_Rec709_BT1886.cube

RED_Achromic_Rec2020_N-Log_to_Rec709_BT1886.cube

Without testing, my assumption is that, in order to take advantage of the .R3D filename hack and work with LUTs, you'd then need to use the RED LUTs. If you want to use the Nikon+RED LUTs, you need to stay in .NEV. You can't both get the "advantages" of the .R3D filename hack, and use the Nikon LUTs. But then, given the filenames above, you likely don't need to.

I will say... looking at the same clips, if you're just wanting to have a really nice baseline to work from, you can either go .R3D filename hack, left to the defaults, and then color grade to your heart's content, or for my tastes, stick with .NEV and run the FilmBias LUT. Both are fantastic starting points, and I have to do very little in the way of correction to get them looking nice. I think the bonuses you pick up going with the .R3D filename hack aren't really worth the hoops you need to jump through, though I did just now save some color management presets to make this a no-brainer going forward.

For reference, here are Nikon's color management settings for their LUTs - this is from Nikon:



And in comparison, these are the settings for working with .NEV files renamed to .R3D:



Alternatively you can work in the Davinci Wide Gamut Intermediate timeline color space and it works equally well:



--
http://jimlafferty.com
Evocative beats academic.

 

[patrol_taking_9j]

RAW is RAW is RAW. There is no color information yet, and said RAW footage can be transformed to whichever color space you desire. If you wanted to use ARRI log or any other gamma and color space, you can do so from RAW footage any camera capable of recording RAW.

So the trick of renaming NEV to R3D files simply allows access to the much improved IPP2 color workflow, ISO and exposure strategy of R3D RAW, as well as the better chroma noise reduction, highlight roll offs, and so on. (This is at the RAW stage.) Then, we take the output of the RAW stage and put it into a gamut and gamma of our choice. I'd recommend using RWG/LOG3G10, because as was pointing out earlier in the thread there are a wide variety of creative LUT's available directly from RED. Some of which are the obvious inspiration for the Nikon/RED luts, like FilmBias, FilmBias Offset, and so on.

You have a choice here - RED recommends using the creative LUT's for creative effect only, as long as your NLE is IPP2 compliant, and not using the LUT's for both creative effect and conversion to a delivery color space and gamma. So in the LUT zip from R3D, don't use the conversion luts to rec 709. You can easily handle the conversion using a CST in resolve, and result will be the same, but better and more flexible because you will apply the LUT effects in RWG and not burn in 709.

When processing N-RAW renamed to R3d, first use Camera RAW to output to RWG/LOG3G10. I would use 3 nodes in resolve. the 1st node where I apply the creative R3D LUTs (not the nikon luts) or any other effect. Next is a CST to take input from RWG/LOG3G10 and output DWG/D-Intermediate, and then a 3rd node to take DWG/Intermediate and go out to Rec709/2.4 or whatever you want.

On the ZR, I am having a hard time seeing how dynamic range will be any different compared to z6iii when filming in RAW. As long as both cameras are exposed the same, you can take the RAW output from either camera and they should be identical. Of course if you film and process the z6iii in say h.265 n-log and film and process the Zr in h.265 RWG/LOG3G10, the output will be different because the two log curves differ; but I suspect if you take the z6iii RAW outputs and process into RWG/LOG3G10 it'll be exactly the same as the Zr.

It's important to separate the two steps in the pipeline. There's the capturing of the RAW sensor data, then outputting that RAW data and interpreting it into presumably a LOG space.

 

[jlafferty]

RAW is RAW is RAW. There is no color information yet, and said RAW footage can be transformed to whichever color space you desire.

Yep, I know. This is a point made in one of the videos I shared earlier. However, setting Color Science to Davinci YRGB Color Managed reads camera metadata and assigns color ingest values, so it will affect your footage.

If you wanted to use ARRI log or any other gamma and color space, you can do so from RAW footage any camera capable of recording RAW.

Yup.

So the trick of renaming NEV to R3D files simply allows access to the much improved IPP2 color workflow, ISO and exposure strategy of R3D RAW, as well as the better chroma noise reduction, highlight roll offs, and so on.

Yup.

I would use 3 nodes in resolve. the 1st node where I apply the creative R3D LUTs (not the nikon luts) or any other effect. Next is a CST to take input from RWG/LOG3G10 and output DWG/D-Intermediate, and then a 3rd node to take DWG/Intermediate and go out to Rec709/2.4 or whatever you want.

Everything I've read is creative LUTs go last in a node sequence, obviously with the exception of an output CST if you need that. So, assuming project settings are as above, I just do an empty node for primaries, followed by a second node that takes the LUT.

RED provides IPP2 LUTs that are solo or have an embedded output color transform - if you keep your project set to what I described above, working with the .R3D renamed files, and simply drop an IPP2 solo LUT on your footage, things look as expected.

I'm just not yet convinced that it's meaningfully different - and I emphasize for my work and tastes - to change filenames, jump into the IPP2 workflow, drop the same LUT I would on an NEV file as is (FilmBias). I'm not seeing a useful difference in output quality. I haven't had to battle blue channel clipping or hit my work with a lot of noise reduction yet, and when I do maybe the IPP2 workflow will be invaluable.

OTOH I will say... if I stick with NEV and add the Nikon+RED FilmBias LUT, or if I go with the R3D filename change, and leave the footage at their defaults, I do prefer the look of the R3D renamed footage untouched by a LUT :-|

On the ZR, I am having a hard time seeing how dynamic range will be any different compared to z6iii when filming in RAW. As long as both cameras are exposed the same, you can take the RAW output from either camera and they should be identical. Of course if you film and process the z6iii in say h.265 n-log and film and process the Zr in h.265 RWG/LOG3G10, the output will be different because the two log curves differ; but I suspect if you take the z6iii RAW outputs and process into RWG/LOG3G10 it'll be exactly the same as the Zr.

It's important to separate the two steps in the pipeline. There's the capturing of the RAW sensor data, then outputting that RAW data and interpreting it into presumably a LOG space.

Yeah, I agree on all points you're making here, emphasis on suspect.

--
http://jimlafferty.com
Evocative beats academic.

 

[NickZ2016]

Given the first LUT I wish to apply is "N-Log_BT2020_to_REC709_BT1886_size_33" in the COLOR workspace I add 3 nodes with CST -- input to Rec.2020 N-Log; then Rec 2020 to DaVinci Wide Gamut/DaVinci Intermediate and finally DaVinci Wide Gamut/DaVinci Intermediate to Rec.709 Gamma 2.2.

Colour managed if you just click auto will take you out of LOG.

What those three nodes are doing is ending up in 709 with three steps. Editing inside 709 is not the best place to be.

The Red film bias LUT is expecting NLog inside 2020. You're out of 2020 and out of Nlog.

If you want to add the Nikon/Red LUTs turn off auto colour managed.

Very first node you could add RED_FilmBias_Rec2020_N-Log_to_Rec709 which sticks you into 709 but at least you're give the LUT Nlog that it expects. I think that would work.

Either that or turn auto back on. Add a node that takes the gamma back to the Nlog the Red Lut is expecting . Do this at the end.

 

[patrol_taking_9j]

Yes - or convert to R3D and use the actual RED LUT's, not the Nikon RED luts.

One downside is that in R3D you lose the lens corrections, but gain ISO control and Chroma noise reduction. Also control over tone mapping and highlight roll off behavior.

I shot a large series of back to back shots comparing N-RAW, N-RAW R3D, ProRes RAW, ProRes 422 HQ, and H.265. Honestly, they are practically indistinguishable. The biggest benefits come from shooting and exposing for N-LOG. Going to N-LOG drops all the extra sharpening and everything in H.265/ProRes, so remember to compensate for that.

I like the greater controls you can access with the R3D workflow inside resolve though. It feels like Nikon could pretty easily fix a few small things and make N-RAW a 1st class citizen, so I hope they do that soon.

Otherwise, ProRes 422 HQ in either SDR or N-LOG is stellar, and H.265 is right there. But at the end of the day I would like to shoot everything in N-RAW, because the file sizes are way more manageable, and it preserves the most content, but is currently let down by shortcomings in the SDK, and the R3D conversion while nice is missing those lens distortion corrections still.

Re: The luts/nodes/cst, if you're using the solo creative luts, I would set the R3D clips in camera raw to be RWG/LOG3G10, then make the 1st node your effects node where you do your color adjustments (still in RWG/log3g10), then 2nd node the solo lut. It is expecting RWG/LOG3G10, so no problems there. Then you can take a 3rd node to convert out to whatever delivery format you want. But there are many ways to skin that cat. Just be sure to be aware of your clip colorspace and gamma and your timeline colorspace and gamma. I don't think you're giving anything up by making your color edits in RWG.

Otherwise if you're sticking N-RAW the Nikon luts expect Rec.2020 N-LOG and output Rec 709/2.4(bt.1886). So if you're using those I would do everything in 2020/nlog.

 

[ClintGryke]

Yes - or convert to R3D and use the actual RED LUT's, not the Nikon RED luts.

One downside is that in R3D you lose the lens corrections, but gain ISO control and Chroma noise reduction. Also control over tone mapping and highlight roll off behavior.

I shot a large series of back to back shots comparing N-RAW, N-RAW R3D, ProRes RAW, ProRes 422 HQ, and H.265. Honestly, they are practically indistinguishable. The biggest benefits come from shooting and exposing for N-LOG. Going to N-LOG drops all the extra sharpening and everything in H.265/ProRes, so remember to compensate for that.

I like the greater controls you can access with the R3D workflow inside resolve though. It feels like Nikon could pretty easily fix a few small things and make N-RAW a 1st class citizen, so I hope they do that soon.

Otherwise, ProRes 422 HQ in either SDR or N-LOG is stellar, and H.265 is right there. But at the end of the day I would like to shoot everything in N-RAW, because the file sizes are way more manageable, and it preserves the most content, but is currently let down by shortcomings in the SDK, and the R3D conversion while nice is missing those lens distortion corrections still.

Firstly and importantly, you are not actually converting the file type, just the extension, and you are still working with N-RAW but in an IPP2 pipeline. Wait for the ZR and then talk about R3D. In the meantime, it's a processing hack in Resolve

Secondly and also importantly, you are not mentioning any of the actual benefits of processing as N-RAW in Resolve. With N-RAW there are a number of controls exposed that are not available in IPP2 and some of these controls are incredibly useful. in particular, the Highlights and Shadows controls are very useful for recovering highlight and shadow detail (as long as there is no clipping) and these are absent in IPP2. The lack of ISO control in the normal N-Raw pipeline is neither here nor there as the exposure control does exactly the same thing. I have not tested the Chrominance Noise Reduction control but I question how useful this is without a Luminance Noise Reduction control as well. If I am going to need noise reduction at all, I will want to apply both in any case using Resolve's NR or Neat Noise.

So I think it is a great exaggeration to say that N-RAW is treated as a second class citizen. I am more than happy with it and moreover, because of the way I process in Resolve, I have never come across the blue channel clipping issue which is Resolve-specific in any case and nothing to do with Nikon. IPP2 and normal N-RAW processing in Resolve are a little different for sure but the jury is out for me in making judgements.

Thirdly, you may have shot a large series of shots in various file formats and tone modes but you are missing some very vital information if you think that "Honestly, they are practically indistinguishable." This will depend strongly not only on what you shot but also on how deeply you examine the resulting clips. It will also depend on the camera used which you don't actually mention, as well as the monitor used to view the clips. You also don't mention anything about frame sizes. For example, there are significant differences between some flavours of N-RAW and ProRes Raw with the Z8 and differences between the different flavours of ProRes Raw itself. There are very significant differences in shadow areas between raw and non-raw codecs, especially if shooting in SDR tone mode. And there are very significant differences, particularly in the shadows, between N-Log and SDR tone modes for ProRes and H.265.

Internal Nikon H.265 is 10-bit 4:2:0 only, which may not be generally noticeable unless you look for it, but there will certainly be differences between that and 12-bit raw. I notice differences in the richness of colour of the same scenes between H.265 and N-RAW. In addition, I've seen H.265 glitching when shooting backlit scenes where the same scenes in N-RAW were glitch-free. I assume this is down to differences in bitrate. Banding in blue sky may also occur in 10-bit but not in 12-bit.

So, in my experience, great care is required in making sweeping generalisations that don't hold up to scrutiny.

 

[jlafferty]

you may have shot a large series of shots in various file formats and tone modes but you are missing some very vital information if you think that "Honestly, they are practically indistinguishable." This will depend strongly not only on what you shot but also on how deeply you examine the resulting clips...

Internal Nikon H.265 is 10-bit 4:2:0 only, which may not be generally noticeable unless you look for it, but there will certainly be differences between that and 12-bit raw. I notice differences in the richness of colour of the same scenes between H.265 and N-RAW. In addition, I've seen H.265 glitching when shooting backlit scenes where the same scenes in N-RAW were glitch-free. I assume this is down to differences in bitrate. Banding in blue sky may also occur in 10-bit but not in 12-bit.

So, in my experience, great care is required in making sweeping generalisations that don't hold up to scrutiny.

I appreciate all that you've written here. I think it's fair to say "with regard to codecs, proceed with caution and test where you can ahead of shooting, using conditions as close as you can to your final shooting scenario to see what problems exist." Or... "H265 has limitations and they're made most obvious in high contrast scenes with a lot of highlight areas in the frame."

OTOH, for a lot of what many people, myself included, are shooting, when properly handled, 4:2:0 H265 in Nlog can be really nice, if not outright beautiful. I do prefer Nraw, but the file sizes are just insane for 6k60, even 6k24.

 

[ClintGryke]

you may have shot a large series of shots in various file formats and tone modes but you are missing some very vital information if you think that "Honestly, they are practically indistinguishable." This will depend strongly not only on what you shot but also on how deeply you examine the resulting clips...

Internal Nikon H.265 is 10-bit 4:2:0 only, which may not be generally noticeable unless you look for it, but there will certainly be differences between that and 12-bit raw. I notice differences in the richness of colour of the same scenes between H.265 and N-RAW. In addition, I've seen H.265 glitching when shooting backlit scenes where the same scenes in N-RAW were glitch-free. I assume this is down to differences in bitrate. Banding in blue sky may also occur in 10-bit but not in 12-bit.

So, in my experience, great care is required in making sweeping generalisations that don't hold up to scrutiny.

I appreciate all that you've written here. I think it's fair to say "with regard to codecs, proceed with caution and test where you can ahead of shooting, using conditions as close as you can to your final shooting scenario to see what problems exist." Or... "H265 has limitations and they're made most obvious in high contrast scenes with a lot of highlight areas in the frame."

OTOH, for a lot of what many people, myself included, are shooting, when properly handled, 4:2:0 H265 in Nlog can be really nice, if not outright beautiful. I do prefer Nraw, but the file sizes are just insane for 6k60, even 6k24.

Yeah sure I agree. My points were really around claims that the various codecs are practically indistinguishable - they're definitely not. I wasn't at all saying that Nikon H.265 is no good, it's actually very good and I use it without any concerns. It depends on what I'm shooting whether I use N-RAW or H.265. File size is definitely a consideration.

 

[Tom Stites]

you may have shot a large series of shots in various file formats and tone modes but you are missing some very vital information if you think that "Honestly, they are practically indistinguishable." This will depend strongly not only on what you shot but also on how deeply you examine the resulting clips...

Internal Nikon H.265 is 10-bit 4:2:0 only, which may not be generally noticeable unless you look for it, but there will certainly be differences between that and 12-bit raw. I notice differences in the richness of colour of the same scenes between H.265 and N-RAW. In addition, I've seen H.265 glitching when shooting backlit scenes where the same scenes in N-RAW were glitch-free. I assume this is down to differences in bitrate. Banding in blue sky may also occur in 10-bit but not in 12-bit.

So, in my experience, great care is required in making sweeping generalisations that don't hold up to scrutiny.

I appreciate all that you've written here. I think it's fair to say "with regard to codecs, proceed with caution and test where you can ahead of shooting, using conditions as close as you can to your final shooting scenario to see what problems exist." Or... "H265 has limitations and they're made most obvious in high contrast scenes with a lot of highlight areas in the frame."

OTOH, for a lot of what many people, myself included, are shooting, when properly handled, 4:2:0 H265 in Nlog can be really nice, if not outright beautiful. I do prefer Nraw, but the file sizes are just insane for 6k60, even 6k24.

Both these comments touch on something that many people in this digital age seem afraid of...testing, testing, testing...or shooting, shooting, shooting. And this seems to be worse with the still shooters than video shooters. You'd think we are still back in the day when testing meant spending real money for film and processing...often lots of it!

To a person, those involved in this thread are actively testing, learning and sharing what works, what doesn't and when you might use one mode over another...instead of coming on a forum and expecting/getting some general answer that applies to one situation and thinking/hoping you're all set.

Each time you're out shooting, whether testing or whatever, you're exposing yourself to the endless variety of conditions that you MUST learn to recognize and compensate for

When I was a kid growing up in rural Central Florida and a member of a 4-H club, one of our mottos was "Learning by doing"...simple, but incredibly powerful. Especially with digital photography and videography, it's a motto to think about every day. It's kind of the old-fashioned version of FAFO!

Cheers!