Better LUT for good skin tones - for Odyssey 7Q+ & FS700 ?

Thanks Jonathan, and thanks everyone for the feedback. I'm glad these new options are working for people and I'm eager to hear some feedback after using either the JW or Betz LUTs in the field.

I'll create an updated version soon with slightly less contrast and saturation, and I'll try to match the exposure more closely to the CD LUT.
 
Mitch or Nick, I wonder if you could provide any insights into how you mathematically shifted the exposure of the Sony LUTs to create your exposure compensation LUTs. I would love to do something similar to accurately create a -1 LUT from my current -2 LUT. Thanks much.
 
jonbetzfilms said:
Mitch or Nick, I wonder if you could provide any insights into how you mathematically shifted the exposure of the Sony LUTs to create your exposure compensation LUTs. I would love to do something similar to accurately create a -1 LUT from my current -2 LUT. Thanks much.
Click to expand...
I use my own Python scripts to perform mathematical calculations on the values in the LUTs, based on the equations Sony publish in their S-Log white papers. But I am afraid I have also written my own interpolation functions, so the detail is more complex than I can go into in a forum post!
 
jonbetzfilms said:
Thanks Jonathan, and thanks everyone for the feedback. I'm glad these new options are working for people and I'm eager to hear some feedback after using either the JW or Betz LUTs in the field.

I'll create an updated version soon with slightly less contrast and saturation, and I'll try to match the exposure more closely to the CD LUT.
Click to expand...

Tried your LUT on a test shoot last night and I would say it's definitely over saturated and under exposed. But the colours are great!
 
Nick_Shaw said:
I use my own Python scripts to perform mathematical calculations on the values in the LUTs, based on the equations Sony publish in their S-Log white papers. But I am afraid I have also written my own interpolation functions, so the detail is more complex than I can go into in a forum post!
Click to expand...

Any chance you'd be willing to share your code? It would be quite helpful to develop a simple app to transform exposure by a given stop amount as we generate more LUTs. I would be happy to help with the coding. Anyway, it would be great to see what you've accomplished if you're into the whole open source thing :)

pythonman said:
Tried your LUT on a test shoot last night and I would say it's definitely over saturated and under exposed. But the colours are great!
Click to expand...

Great to know. Thanks Jonathan. I will tweak the LUT and re-upload after Thanksgiving.
 
jonbetzfilms said:
Mitch or Nick, I wonder if you could provide any insights into how you mathematically shifted the exposure of the Sony LUTs to create your exposure compensation LUTs. I would love to do something similar to accurately create a -1 LUT from my current -2 LUT. Thanks much.
Click to expand...
you don't need mickshaw's algorithmns. you can do it in cameramanbens LUTCALC, which is free software.
 
Here are links to two updated exposure compensation LUTs for use with the FS700/Odyssey 7Q.

- same exposure shift as Convergent Designs preset exposure compensation LUTs (SONY_EE_SL2_L709A-1 and SONY_EE_SL2_L709A-2)
- improved color accuracy

https://www.dropbox.com/s/8gz1sg5bqlpyvs6/Betz_SL2-LC709A-1.cube?dl=0
https://www.dropbox.com/s/88agwpwjc720wup/Betz_SL2-LC709A-2.cube?dl=0

Method for creation:

1. I started by creating the following basic 3D LUTs in the LUTCalc program:
Camera: Generic
Stop Correction: -1.2 or -2.2 (the additional 0.2 stop correction was needed to match CD's exposure compensation LUT's)
Rec Gamma: S-Log2
Rec Gamut: Passthrough
Out Gamma: LC709A
Out Gamut: Passthrough
Input Range: Data
Output Range: Legal
Grading LUT

2. In Resolve I applied the newly created LUT to the test clip and reduced the contrast slightly to match the contrast range of CD's LUT's by raising/lowering the bottom and top points of the luminance curve.
3. Saturation increased from 50 to 65 in Resolve
4. In a new node I applied the auto correction for the SpyderCheckr 24 using the default settings (all Rec.709 and target color temp 6500)

As always, feedback is always appreciated.
 
Awesome, thanks Joe. Just did a fashion shoot today, so will test your LUTs alongside mine tomorrow in the edit, and will report back. I took still photos with my Canon 7D for colour reference of the fashion items in the shoot, the 7D's photos looked realistic on set when placed side-by-side with the actual clothing next to the photos on my onset monitor, so it should be a good real world test.
 
First impressions very good Jon. Although I've only tested in one set of conditions, I'd say it's probably the best yet!

The reds are much, much better, certainly better than with the LUTs I created. The blues are still not quite perfect, but the best I've seen so far with Slog2 from the FS700/O7Q combo.

I'll post some pics shortly to show comparison.

Thanks Jon for your time creating these, really appreciate it.
 
Great to hear. I look forward to seeing your samples.

I did not do any Hue vs Hue or Hue vs Sat corrections for this LUT, but could easily lower the red saturation or shift the blue hues if needed. I ordered my own Spydercheckr so will do a more accurate assessment of the color reproduction next week when my 7Q is back from service.
 
OK, well this is not a scientific test by any stretch of the imagination. But on Friday I did a shopping-channel-style shoot for a client, where I shot about 200 small baby clothing items being demonstrated.

I had my FS700/O7Q+ shooting close-ups of the set, whilst just sat in front and slightly below the FS700 I had a Canon 7D on a tripod to take separate still images - see attached image of the shoot arrangement taken on my phone!

JW on Set Dec 2015 smaller.jpg

The photos I took on the Canon 7D were taken on my Macbook Pro through tethering in Lightroom. This way we could physically compare the photos to the actual item. By holding the clothing item next to the Macbook Pro's screen I would say the colour accuracy of the 7D's images were as near as dammit close to the real thing - very accurate in other words! I do accept this is not scientific, realising there are other factors including the calibration of the Macbook Pro's screen etc, but to me this is a real-world test that I find useful.

So on that basis see below screen grabs of one example, with the various LUTs showing comparison, plus the 7D photo. I shot in SLOG2 on the FS700 and was fully white balanced. Also note that I have applied a very slight increase in contrast in Premiere post adding the LUTs (I did this on the first one, so had to do so on the rest for consistency), but no changes to the colours were made.

I note particularly that the blues are not accurate with any of the LUTs compared to the 7D's photo - which like i said reflected very well the real thing.

Note, the "NEW SLOG2 LUT Resolve -2" was the latest LUT I created as linked in post #20 within this discussion thread.

Let me know what you think.

SLOG2 Image - No LUT.jpgSONY_EE_SL2-L709A-2.jpgNEW SLOG2 LUT Resolve -2 (JW).jpgBetz_SL2-LC709A-2.jpgCanon 7D - Standard PP.jpg
 
Last edited:
Thanks for putting this together Jonathan. Looks great! Of course my first reaction is, how did I use that original Sony_EE_SL2_L709A-2 LUT for so long when it so clearly mismanages the colors of the FS700 (as it was designed by Sony for a different camera with a different color output)? Glad we are making progress here, and hopefully lots of FS700 owners out there see this thread and start using these new LUTs to get the best images from the FS700.

I like your process here and I'll do something similar when I get my Spydercheckr. It should be easy to tweak the blues a bit so we get even closer. The important conclusion here is that the FS700 does offer a nice color palette. We just needed the right LUT to bring those colors out. This should be a game-changer for FS700/Odyssey users still using the CD LUTs.
 
On my list for things to check/fix:

- balance saturation by hues (at first glance it seems like the greens may be a hair desaturated in my LUT in your tests, but I will do a more accurate saturation balance with the Spydercheckr).

- adjust the blue hues slightly toward violet.
 
This is exciting, can't wait for this to be done!! Might be a silly question, but once this lut is perfected, will this be better or equal to slog3? Or are they their still clear benefits to slog 3 over slog 2 with a good lut?
 
Thanks Jon. Yes as you say, I would wait to do your hue vs hue adjustments using the SpyderCheckr card rather using my above examples, but we're on the right track for sure! One other request...it might be a nice idea to have a LUT that purely corrects the colours only, leaving it as a SLOG-2 image, that leaves us free to do other types of grade other than REC709. Look forward to your next versions Jon!!

Also one for Mitch if you are reading this - one thing that struck me, that I've mentioned before, is that the FS700 (in my experience) captures the most accurate colours using NO Picture Profile at all, albeit with less dynamic range. What would be very nice, especially for controlled lighting situations as above, is for the Odyssey to input the Raw signal and record in ProRes with ANY or with NO Picture Profile, and not just be limited to SLOG2 or REC709 800% (even this profile has inaccurate colours). This would also allow cleaner images where appropriate, rather than the noisy ISO2000 SLOG2 images! Would this be possible for a future update? Thank you.
 
pythonman said:
Also one for Mitch if you are reading this - one thing that struck me, that I've mentioned before, is that the FS700 (in my experience) captures the most accurate colours using NO Picture Profile at all, albeit with less dynamic range. What would be very nice, especially for controlled lighting situations as above, is for the Odyssey to input the Raw signal and record in ProRes with ANY or with NO Picture Profile, and not just be limited to SLOG2 or REC709 800% (even this profile has inaccurate colours).
Click to expand...
Rec.709 (800%) in the Odyssey is, like the same profile in the camera, simply the camera's native colour space with a 1D LUT (curve) applied. There is no colour matrix, or other attempt to create "accurate" colours.

I can't speak for what options Convergent Design may offer in future, but having worked with them on analysing the camera processing of the FS700 and FS7 to help create the options that are currently available in the Odyssey, I can tell you that replicating what the camera does is certainly not trivial. And the various picture profiles (and I suspect also "no profile") still have a fair amount of processing going on, much of which is Sony "secret sauce", and not necessarily easy, or maybe even possible, to replicate from the Raw output of the camera.
 
You must log in or register to reply here.
Back
Top