Gamma/Knee problems: How do I correct this?

[Matt Lyon]

Hi Tim,

Can you elaborate more on what you mean by “it displays incorrectly”?

Also, have you tried setting your system to be fully 2.2 gamma. You may get more predictable results: Set your display’s gamma to 2.2 in the control panel. Set Photoshop to work in sRGB color space, and embed your images with the same.

sRGB is fairly close to rec 709 color space, so it is a good starting point for still image creation.

Matt Lyon
Editor
Toronto

[Tim Ward]

Thanks for the reply Matt, but I think we’re talking a bit of apples and oranges here.

[Matt Lyon] “Can you elaborate more on what you mean by “it displays incorrectly”?”

Sorry, I didn’t write that clear enough. I’m referring to the ramp I create in PS, then import into FCP. The ramp should be completely straight. Interpreting as “Source” or “1.8” gamma brings the waveform closest to correct.

[Matt Lyon] “Also, have you tried setting your system to be fully 2.2 gamma. You may get more predictable results: Set your display’s gamma to 2.2 in the control panel. Set Photoshop to work in sRGB color space, and embed your images with the same.”

That will only affect how my computer display looks. The Preview and Canvas displays don’t tell me how it will look on a video display – they are only approximations. Yes, I work in sRGB since viewing it in PS approximates how it’s viewed in video, but FCP doesn’t read color profiles. Using TIFFs, Adobe RGB, sRGB, and no profile embedded show the EXACT SAME waveform (as above), and they wouldn’t if FCP used the profiles.

I’ve re-created this on my home system with the same specs as above, except it’s a Mac Pro 1,1 with QT 7.6.4. Hardware scopes on both systems read the same as the image above. I don’t really care if the Canvas gamma is off – I DO care if it is off where it matters: on my video outputs.

[Matt Lyon]

okay, I see where you’re coming from Tim. Yes, FCP doesn’t read color profiles, but it does read the embedded gamma flag. To be honest, I’m not sure how that flag is set to begin with. I thought it was part of the color profile, but I guess I’m wrong.

But I don’t think a bad gamma flag would account for the crazy toe/shoulder problem on your waveform readout.

Are you sure Photoshop is actually spitting out a perfectly linear ramp?

I did a little test here, making a ramp in GIMP, and it shows as perfectly linear on my software waveform monitor in FCP (I know, I know — the software scopes aren’t telling the full story).

Sorry I can’t be of more help, I don’t have Photoshop here to test with.

Matt Lyon
Editor
Toronto

[Tim Ward]

[Matt Lyon] “Are you sure Photoshop is actually spitting out a perfectly linear ramp?”

Doh! Didn’t think of that. The problem was indeed in PS (caused by me)…with assistance from FCP. A TRUE gradient in PS is created with 0% smoothness, not 100% as it is set at default. Created a new ramp, exported .tif, imported in FCP, and after setting the gamma to 2.2 for the ramp, it became flat (Source chose 1.8). Great! I also used FCP’s ramp generator for comparison, sending it through PS and back again, it looked fine, but THIS .tif needed gamma set at 1.8/Source. Why? Good question. Possibly because of the way PS interpreted it upon opening it. Comparisons within PS did show a difference between the two.

Ultimately, the reason I started this whole thing was to make sure I was correctly interpreting graphics used within FCP.

Thanks a lot for the help, Matt!

[Matt Lyon]

Glad to help and I’m happy you sorted it out. For whatever reason, my test image from GIMP needed to be set to 1.8 to display properly. But picking the “source” gamma option worked too. I’m not sure what the rhyme or reason is behind the inconsistencies we are seeing … But I guess our different results are a good example why ever user should do their own tests!

Matt Lyon
Editor
Toronto