Artifact in DPXs output from RedCine

[N. Keller]

Hi all,

I just saw some test footage shot on the Red at 4K and transferred to 35mm. There were distinct artifacts in out of focus areas of all shots. You could see clearly defined bans of "steps" from light to dark, instead of a smooth transition. The type of thing you would see in very reduced bit-rate images, or badly compressed digital images, especially in shadows. In this case it was happening into the midtones and highlights, but ONLY in out of focus areas.

I was told the steps from shooting to printing to film were ONLY to take the .R3D files and convert them to 2K DPXs in RedCine. No color correction. I can only imagine something went awry here. I'm going to look into the exact process that was followed, but since I've never worked with Red footage or RedCine, I was wondering if anyone had any idea how this artifact could have been generated? Is there any chance it's something that happens during camera capture? I wonder if perhaps the DPXs were accidentally created from the QT proxies? Is that even possible?

Since I can't upload an actual image, I've done my best to emulate what I observed with a picture in Photoshop. This is only an approximation. What was interesting was the fact that it could only be observed in out of focus areas. Any insight would be of great help!

Thanks!

 

[Chris Swinbanks]

Quantization artifacts on print should only be apparent if the data has been "mistreated", and I would presume its got to do with limiting the bit depth or overly compressing the data at some point, as you've suggested. You'd really need to provide much more info about the conversion process.
My first suspect would be that perhaps its been limited to a 685 cv range.... please post what you find out about it.....

 

[N. Keller]

Quantization artifacts on print should only be apparent if the data has been "mistreated", and I would presume its got to do with limiting the bit depth or overly compressing the data at some point, as you've suggested. You'd really need to provide much more info about the conversion process.
My first suspect would be that perhaps its been limited to a 685 cv range.... please post what you find out about it.....

Thanks for the response--and the term for that type of artifact! I forgot to mention that the artifact was extremely visible in all the out of focus areas of the DPXs on the DI system, but only noticeable in one shot of the 35mm print.

I will post again once I find out what happened exactly.

 

[Graeme Nattress]

Is it possible that files were exported as linear light 10bit DPX files from Redcine, as that would surely cause visible posterisation?

Graeme

 

[N. Keller]

Is it possible that files were exported as linear light 10bit DPX files from Redcine, as that would surely cause visible posterisation?

Graeme

Aaah... It's possible. The people who output the DPXs were the people renting the RED, and they've never done film out work before, so it could be they didn't know what they were doing. Thanks SO much, as this points me in the right direction. I'll be going in this week to pay them a visit and find out exactly what they did and will report back.

Nayeli

 

[Graeme Nattress]

Well, without an R3D to do a test conversion here, to actually look at the data, that's the best guess I can give. I hope it points you in the right direction though.

Graeme

 

[N. Keller]

Well, without an R3D to do a test conversion here, to actually look at the data, that's the best guess I can give. I hope it points you in the right direction though.

Graeme

Thanks so much. I'll see if it's possible to get an R3D to test. Can it be just one frame? I'll output my own tests as well. Is there any feedback on 10 bit log DPXs being the best for film out, by the way?

Nayeli

 

[Graeme Nattress]

Sure, as long as that one frame allows me to see the issue. I've heard of some good filmouts via 10bit DPX.

Graeme

 

[Chris Swinbanks]

I've heard of some good filmouts via 10bit DPX.

Graeme

speaking of which, Graeme... did you receive the email about the a) transcode & b) NR test frames on ftp I sent you? Really need to get some clarification on this as quick as possible. cheers Chris

 

[Graeme Nattress]

Yeah, I've been looking at the stuff. I'll email you.

Graeme