Messages

ISO is a physical thing... you can't use iso 100 and then try to "recover" data that would be in ISO 1250, etc.

Okay, thanks. So because the ISO changes the sensor's actual sensitivity, it gets recorded into the raw file? And yeah the memory did go down.

I'm not sure if this is supposed to happen, but the ISO is being baked into the raw files and I can't recover any detail at all, even when correctly exposed. I know that I should be able to recover highlights because ISO is supposed to be metadata, but when I look on camera at ISO 80, there is detail in the highlights, but when I look at ISO 320 or 640, there will be detail in the shadows but none in the highlights. But after I run raw2dng and bring the files into ACR, it's a no go. Can anyone chime in on this?

very useful information. thanks!
so basically if you're shooting raw, use cineon.
if you're upconverting, leave it at video gamma.
as for picture styles, i've been using visioncolor and lightform

Glad to help! Not sure if I mentioned this earlier, but for raw you could use the CF databases to have a preset look to help for editing and a cineon database for grading after the edit is finished. And yeah since you're using regular video gamma profiles, stick to video gamma.

Can you tell me please: if i render my DNG file in After Effects to cineform 4-4-4 : it will be raw? or just compressed 12 bit 4-4-4?

It'll be 444 if it's coming out of After Effects, not raw. The only way to get CF Raw at the moment is through Studio Premium/Pro or the command line tool dpx2cf that's included with CF.

i've read to leave it to protunes.
what are the pros cons on protunes vs cineon?
coming from 8-bit to 10bit?

You don't have to leave it on Protune. Protune set as the output curve will give you a "finished" look. When working in 10-bit and 12-bit, especially for color grading, cineon (logarithmic gamma curve), compresses all of the data into a range where each stop of light in the file gets around the same number of code words (bits of data for detail). This is why film is scanned to cineon log files. The entire dynamic range of the film frame can be captured and stored in the log file. Same principle applies to raw. That's why ARRI, Red, Sony, Blackmagic Design, etc. have there own versions of cineon log designed specifically for the sensors of their cameras. If you compare Alexa Log-C, RedLogFilm, and BMD Film (I actually did a post on BMCUser about using Cineon 95-685 to get extremely close to this curve with CF) , you'll see that they are all fairly similar looking with differences mainly in color science. With Cineform we can take a raw file and basically transform it to a mathematically correct cineon file, Log-C file, S-Log, or whichever curve you choose file. However, when compositing or editing, I'd work with the Protune output curve so you don't have to create proxies and you can stay with the CF raw file until you actually have to render an intermediate composite or something like that.

For up-converting the 8-bit to 10-bit, it depends largely on which profile you shot with. If you used Cinestyle, there is a preset in CF Studio to set the i/o curve to CStyle. For any other profile I'd say leave it at Video Gamma because in the H.264 file there's only about 6-8 stops of usable dynamic range depending on the ISO setting you shot with.

i've been using cineform studio premium for several weeks now for a short film we've been working on.
converting 5d mark ii h264 to 10-bit cineform. its great that it gives me much better flexibility in doing aggressive color grading.
and loving the white balance picker. saving me tons of time.

but downsampling from raw to 10-bit cineform isn't as good as raw to dng sequence using after effects.
ACR has a much better control than cineform. and i dont think cineform even has highlight recovery slider.
cineform is great for upconverting from 8-bit to 10-bit color space though.
just my two cents.

Cineform Raw is 12-bit log, which means it can store the same amount of data as 16-bit linear, more than the 14-bit linear that comes from the Canon DSLR sensors. Cineform doesn't need a highlight recovery slider because it automatically restores all the highlights contained in the raw file. For proof look at the default ACR settings vs Protune with no lut. Or set the output curve to Cineon 95-685 or Custom Log 400 to gain back all dynamic range from the file.

I've been reading through all of the pages in this thread and I can't remember seeing anything about this. On the newest 600d build (edmac popin) from 1% it seems as if the ISO is being baked into the dng files. Adjusting exposure in ACR didn't help at all.

Next thing to try now that I'm getting some nice footage is to convert to LOG intermediaries (in AE) for grading in Resolve. No one has mentioned doing that yet. Were still effectively dealing with REC709 images. LOG should be better for extreme grading and with the extra bit depth the color won't suffer. Fun time

VisionColor is developing VisionColor, VisionLog, and CineTech for Adobe Camera Raw and Lightroom.

I'm not sure if this has been mentioned anywhere before because I couldn't find anything after googling it for a good hour and a half. I believe I've stumbled upon a hidden gem inside of GoPro Cineform Studio Premium; pretty sure it can be done in the free version as well. I converted my footage and they came out merged to my extreme surprise and I almost had a heart attack. But I think I know what did the merging though. I brought 720p 48fps ML HDR footage into CF Studio and opened the advanced settings. They essentially look like this:

Frame Size: 1080p
Frame Rate: 24.00
     Maintain Audio Pitch: [ ]
Speed Up: [X]
     1
Frame Blend (Motion Blur): [X] <----- I'm pretty sure this is what merges the frames to make them the pseudo-log files that result...
File Format: AVI
Quality: Film Scan 1
Remember Settings: [ ]

Please try this for yourself and reply with your findings and see if it gives you similar results. It is worlds faster than using GingerHDR EXR export and you get the benefits of the CF codec(like realtime playback in your NLE). One thing I'm not sure of is if CF Studio uses an optical flow algorithm to apply the motion blur (it looked good to me).

UPDATE:
So from further pixel peeping, it looks like CF Studio isn't mapping the exposure difference (like the option in GingerHDR to change from 1,2,3,4,5 EV). I could be wrong, but that's what it looks like.

Cheers,

Eddie