is our system ok to grade for theatrical screening?

[Mike Most]

Well, if you get very good accuracy, then your setup and specs are fine. If you don’t, they aren’t. I know that sounds a bit snarky, but reality is the only real guide here. It really doesn’t matter what others experience is. Either your setup works or it doesn’t. As for the specifics, if the transform in your DCP software didn’t work, they wouldn’t offer it. Same with Calman and the probe. I know you’re looking for some affirmation here, but the fact is that you’re asking questions about things you haven’t tried yet. Try them and ask questions if it doesn’t work.

[Joakim Ziegler]

Generally I agree with the sentiment of “if it works, it works”, but I’d be wary about “if the transform in your DCP software didn’t work, they wouldn’t offer it”.

DoReMi CineAsset, for instance, ships with a Rec.709 to XYZ transform which is definitely and empathically wrong, and has been wrong for ages (they use a Rec.709 encoding curve to decode, instead of a Gamma 2.2-2.6 curve). I’m not sure why it hasn’t been fixed, but it’s a pretty basic thing.

I wouldn’t trust anything or take anything for granted, but in general if it looks good, it’s good.

—
Joakim Ziegler – Postproduction Supervisor

[Pepijn Klijs]

Sorry Mike, but opendcp does a bad job in colortransform. It is even acknowledged and noted by themself on their website or wiki. It’s not good cause it exists.

Editor/Colorist, Amsterdam, The Netherlands
http://www.pepijnklijs.nl

[Terrence Meiczinger]

The XYZ conversion has been tested on a fully-calibrated projector system and measured with a Photo Research PR-655 SpectraScan Spectroradiometer (https://www.photoresearch.com/current/pr655.asp). All colors came out within specification.

Some people have said they have issues with elevated black levels and others have said it’s perfect. In the cases I have looked into issues were due to using 8-bit source images, not using full-range levels, selecting sRGB/REC709 incorrectly or they have been using EasyDCP player to view on an RGB monitor. I’m not saying there isn’t an issue, I’m not a color expert, but I have not had any issues with material I have personally converted.

Having said that, the proper digital cinema workflow is that the images should already be in the XYZ colorspace prior to DCP creation.

All the source for OpenDCP is available, if anybody can suggest an improvement or identify a problem, please feel to let me know. Any bug fixes, reports or suggestions are greatly appreciated.

[Pepijn Klijs]

Hi Terrence,

I think the biggest improvement opendcp can have is an option to specify the source gamma, as it differs from each user in each part of the world.

I was referring to a thread I read on an opendcp forum, where After Effects was recommended over Opendcp for color transform by some of opendcp’s own people.

Editor/Colorist, Amsterdam, The Netherlands
http://www.pepijnklijs.nl

[Terrence Meiczinger]

I’m the sole developer of OpenDCP and what I said and have always said is that you should provide OpenDCP with DCDM (XYZ) graded material, if possible, as that is the proper workflow. You’ll also have more control over variables such as reference white points.

What other gamma’s would you want? There are generally only 3 sources sRGB, REC709, and P3 which have defined gamma values.

Aside from Adobe, do you have experience with applications you feel have a better conversion than OpenDCP?

[Pepijn Klijs]

[Terrence Meiczinger] “What other gamma’s would you want? There are generally only 3 sources sRGB, REC709, and P3 which have defined gamma values.” In Europe many colorists calibrate their monitor to 2,35 or 2,4 for example, cause that’s the Europian spec for broadcast. It would be nice to have that option.

By the way, I did some googling and realized the discussion I was refering to earlier was from 2011, so things might have changed for opendcp. Here’s the article: https://code.google.com/p/opendcp/issues/detail?id=36&can=1&q=dcdm

My own tests showed better results using Assimilate Scratch and DCP Builder. And not just in EasyDCP Player, I have done extensive testing with a local cinema here, which actually also teached me that Easydcp player is a pretty good reference of how the image will look in the theater. You seem to have other experiences with easydcp player?

Pepo

Editor/Colorist, Amsterdam, The Netherlands
http://www.pepijnklijs.nl

[Joakim Ziegler]

Rec.709 does not have a defined gamma. It has a defined encoding curve, but the decoding gamma is non-defined, and the industry standard de facto curves for decoding vary between g2.2 and g2.4.

However, rec.709 material presupposes this decoding gamma for a dim surround (typical TV watching environment) while DCPs presuppose dark surround (movie theater). Thus, it’s often appropriate to adjust gamma even more when converting from one to the other, a “straight” conversion often ends up too dark looking when going from P3 to rec.709, and too punchy when going the other way.

And then there’s the white point. If you go straight through XYZ when converting between rec.709 to P3, you will experience a quite noticable green/magenta shift, and your whites won’t be pure in the destination space. Some LUT generation tools will let you do white point correction to avoid this, so that pure white in P3 (5800K) will end up being pure white in rec.709 (6500K).

Sorry to go on a rant. It’s just that from practical experience, this stuff is not quite as simple as the basic math and specifications make it look.

—
Joakim Ziegler – Postproduction Supervisor

[Terrence Meiczinger]

Sure, REC709 should have a decode gamma relative to the given scene, that’s one reason why color grading should be done prior to DCP creation. There are lots of variables and there is only so much that can be done downstream in the process.

OpenDCP’s main goal has always been to create a DCP from DCDM material for these reasons. However, since many people don’t have access or knowledge to do the XYZ conversion, one was implemented in OpenDCP for convenience. It’s going to be best effort since gamma, white points, etc can’t be determined from the image itself, so the conversion uses the most common values which will be good enough for the majority of cases.

You really can’t compare the results from a dedicated color grading application or workflow. My point is that OpenDCP’s conversion isn’t wrong, so much as there are more optimal workflows.

As for EasyDCP player, it does a decent job, but you can’t use it as an absolute reference. Their XYZ->RGB is going to be matched to their RGB->XYZ conversion, which may or may not be correct. I know that my tests on a reference digital projection system seem to indicate their conversion is overly dark.

In the end, free tools only take you so far. There is place for open source applications and for commercial ones. As always, your mileage may vary.

I’ll look into adding a gamma option in a future version of OpenDCP. Thanks for the feedback.

[Pepijn Klijs]

The gamma variable would really improve opendcp I think. I am asumming that 8 out of 10 ‘bad’ conversions are related to this issue.

By the way, DCP Builder is also free.

Looking forward to see the improvements!

Regards
Pepo

Editor/Colorist, Amsterdam, The Netherlands
http://www.pepijnklijs.nl