source color sRGB and Rec 709

[P]

Hi,
I was wondering, what exactly is the difference between sRGB and Rec 709 settings in OpenDCP?
I was just wondering, when should you use Rec 709 in opendcp?
Yes, when the source is in Rec 709, but when is that?
If I export to tiff, that's RGB, right?
Is Rec 709 setting only for DPX input?

Hopefully someone can explain this to me,
thanks,
P

[Jack Jones Colourist]

Colorspace is pretty much the same - Gamma is different.

I would imagine 709 is g2.4 and sRGB is g2.2.

I'd also go so far as to guess that these are both designed for full range source material (i.e. no legal scaling applied).

Hope that helps and is right - I might be corrected on this one!

[P]

Hi again.

I was making some tests but I am still a bit puzzled.

I used a test pattern tiff from Michael Cinquin Charbon color profiles website.

The gray values on the right are 10%, 30% and 50%.

I made a conversion to XYZ with open DCP, Photoshop and Easy DCP and then measured the grays in Easy DCP player which does the conversion from XYZ to RGB for my monitor:

• Open DCP sRGB:
  
• 12-13%, 30-31% and 49-50%
  
• Open DCP rec 709:
  
• 17-18%, 35-37% and 53-55%
• Photoshop CS6: converting the tiff from Rec 709 to DCDM X'Y'Z'(Gamma2.6) 5900 (by Adobe) and then converting it to j2c in Open DCP with X'Y'Z' unchecked:
  
• 13%, 30-32%, 49-54%
  
• Easy DCP creator:
  
• 10%, 30-31% and 49-50%
  

Open DCP with rec 709 settings is way off.
Open DCP with sRGB settings seems to me the most correct choice, but there is a big difference around 10%. Should I try and compensate that before converting to XYZ?
I was expecting Adobe to give the same results as EasyDCP creator, but the values from Adobe are quite different while the one's from Easy DCP stay almost the same.

Is the conversion of Easy DCP more precise than the one from Adobe and Open DCP or am I missing something here?

 

[terre...@gmail.com]

You are using EasyDCP Player to perform the XYZ->RGB conversion, so certainly I would expect Easy DCP's RGB->XYZ to conversion to match closest. If they have any differences in their calculations, it will cause false errors in the other methods you used.

I've had the calculations in OpenDCP checked by color space professional. In addition, test patterns were created with OpenDCP and projected on a calibrated digital projector. The color levels were measured with a Photo Research PR 655 Spectroradiometer and all measurements were within tolerance. Now, OpenDCP uses look up tables for performance and it could be that there are some boundary areas where errors are exaggerated. Another source of error could be the bit depth conversion. OpenDCP convert's everything to 12-bit, then does the XYZ conversion. Technically, the source should be 16-bit, the XYZ conversion done, then converted to 12-bit.

I'm not a color expert and I really don't know which product is doing things most correctly. I can only speak for the steps I've done to validate OpenDCP's conversion process. The math and logic seem right, whether it really is, I dunno. If somebody can point to the code and say "hey you idiot, that is wrong", I'll happily fix it. :-)

[Stephen van Vuuren]

First of all, I use After Effects for both conversion to XYZ and J2K encoding and can verify 100% that Adobe’s conversion, if done correctly, it’s 100% accurate. I’ve done well over 100 DCPs including my own projected on the best projectors in the world, calibrated by Barco engineers onsite (I have a clip part of Barco demo) and it was beyond spot on – it was remarkable.

 

Easy DCP Player conversion in real-time is decent but no better than Stereoscopic Player – all those are simple LUTs. One thing to be sure is that the monitor and color calibration on your own system is a factor here. XYZ is much larger than sRGB and DCP J2K is 12-bit, most monitors are 8-bit or less. Wide gamut vs sRGB combined with color calibration can cause issues. So be sure you are seeing everything in an accurate, calibrated environment and be aware of your monitors gamut and internal LUT setup.

 

E.G. Adobe is Wide Gamut, multiple monitor aware with full color management system. EasyDCP is not. I don’t consider EasyDCP a calibrated system even though with display port it does support 10-bit color. But I find no reference in the manual it’s looked at monitor ICC profiles or color management software for calculating it’s conversion.

 

I’m not sure your testing methodology is going to give you the correct values and results.

 

However, all that being said, in my testing OpenDCP colorspace conversion is decent but not great – but I’ve only done a couple of basic tests and not with versions .26 or .27.

 

stephen van vuuren

336.202.4777

 

http://www.insaturnsrings.com/

http://www.sv2dcp.com/

http://www.sv2studios.com/

 

A film is – or should be – more like music than like fiction. It should be a progression of moods and feelings. The theme, what’s behind the emotion, the meaning, all that comes later.

–Stanley Kubrick

 

[Terrence]

Stephen has a nice workflow going and pushing the color handling
upstream is definitely the preferred way. If you aren't fortunate
enough to use a program that allows DCI XYZ conversion, like After
Effects, then you can fallback on OpenDCP.

[P]

Hi,
thankyou both for the replies. It's actually what I was hoping to read.
I made quite a few DCPs with OpenDCP that looked really nice in the cinema.
Now I am looking at AE in order to improve my workflow, but as it is a more complex tool it's easy to make mistakes, so I am using all methods I can think of to sort of check and compare my results before I see them in a cinema.