2k DPX RGB export from Da Vinci to EasyDCP

Hi to all,

We have just finished grading a one hour program made of 6 short films. Our suite is Da Vinci Resolve on Mac displayed on a DCI projector NEC NC1200. The projector has different color profiles (HDTV and DCIP3), but we graded in XYZ color space full range images (from Alexa log Prores 444) with a 3d display lut RGB to XYZ. In order to make the DCP, our lab asked us an image sequence from the Resolve in DPX RGB 2048*858 24fps.
So we exported the project without the lut and sent it to the lab. They use EasyDCP.

We are just back from the first DCP screening and everything looked like the gamma was off, midtones very bright, washed out look. Very different from what we graded.

The lab says that Easy DCP doesn't interpret the dpx files. Wich is very odd to me... Or our projector has a wrong gamma setting in its DCI profile... Wich is also odd to me since it has been calibrated twice in two months.

Anyone has bad experience working with easyDCP ?

I had a quick look at the Easy DCP user manual and here is what it says about input data characteristics:

"Easy DCP uses full input range, means e.g. for 10bit 0..1023 codevalues (black is 0, white is 1023).

RGB means a gamma of 2,2 is applied to the linear RGB values.

The color gamut of the RGB values should be sRGB/ITU 709 with D65 whitepoint."

My guess is that Easy DCP interpreted our data as REC 709 footage with a 2.2 gamma and converted it to XYZ 2,6. Which would explain the washed out look of the DCP.

Where do we go from here? Do we need to apply a render lut to solve this?

If you have some advice, I'm ready to listen...


Thanks

Aurélien

Lylo Post-Production
Paris
France

Thank you very much Jack.
We are going to try the TIFF export from Resolve to see how it goes.

Meanwhile, I asked the lab technician to check again our dpx in his Scratch (this lab has a Scratch grading system / Easy DCP). We wanted to check if our program had the same gamma shift on the Scratch and on the faulty DCP (also we needed to rule out any calibration issue). He says that the dpx is read as log by the Scratch. Of course it isn't, since it's our rendered grade straight out from Resolve. And if you check the curve of our dpx it's far from a log curve.
Anyhow the lab guy forced the scratch to read the dpx as lin. And he says that then the film looked like our faulty DCP (washed off, lifted blacks and midtones)... This would confirm to me our issue : wrong gamma interpretation - we graded on a DCI projector with a 2.6 gamma and a rgb to xyz display lut in Resolve. The dpx is read by the Scratch considering it's a 2.2 gamma level, so as Easy DCP. Therefore everything is lifted.

Why does Scratch reads the dpx as a log? There's no way resolve exports log... Anyway we decided to do the TIFF export from Resolve and test it both in Scratch and easy dcp. Tiff is both linear and xyz... So we should know more from there!

But if you guys have any ideas or thoughts...

Thank you very much.

Aurélien

LYLO POST-PRODUCTION
PARIS
FRANCE

Hi,

By default EasyDCP assumes that your material is in REC 709 (it uses a lut to convert to XYZ I think). You can bypass this conversion and use dpx files that are already in XYZ. In your case you should output the dpx files as they were going to the NEC projector from davinci, using the rgb to xyz 3D lut you used.

If your lab uses EasyDCP+ (a version with more features) you can render in RGB and then apply a 3D lut in the program settings to convert to XYZ, but in your case it´s easier to apply it Davinci.

There are demo versions of easyDCP creator and player available online, so if you keep having issues you can create a dcp and preview it using your projector.