I have worked on projects where the DPX renders were a little buggy in spots, or the relative exposure settings were not optimal, and I had my assistant go back and redo a few shots or a few sequences. But we're talking 5 minutes out of a 120-minute film. In most cases, we were trying to get more detail out of highlights, so it was just a question of dropping the levels half a stop or a stop. We still had lots and lots of range to get the precise picture the DP and director wanted, using a custom LUT, and in some cases custom curves and an extra layer or two.
One caveat with DPX (other than which I fully endorse what Marc and Jake are staying) is that 16-bit DPX is (obviously) preferable to 10-bit, but not all apps support it. This being the Adobe forum it's worth noting that After Effects (at least up to CS 5.5 – I've not tested 6.0) only supports 10-bit DPX.
No 16-bit DPX in AE CS6 yet. However OpenEXR works in AE, supports 16 (or 32-bit) and has tighter compression (at 16-bit) than DPX (at 10-bit) in my experience. It's also supported by resolve, and with ProEXR is supported in premiere and photoshop.
I've written an AE plug-in called DPX Plus to read 16-bit DPX in a pinch. You're still limited to 10-bit when writing out of AE though. In general it's probably better to stay away from it if you can.
I'd like to now just how many users are successfully using EXR to grade with in production settings or how often one NEEDS to use 16 bit DPX for grading? Anyone...?
I don't think anyone needs 16-bit dpx with openexr available, but if it wasn't available then for sure.
We are using OpenExr for our vfx workflow with RCX, Premiere and AE on our feature, right now. The post-vfx EXR clips combine with the R3Ds for grading.
I'm sure ILM uses it.
I know at least one Scratch based DI house in LA that prefers to work with 16-bit tiffs. But at 5k they are so dang big (60-80mb/frame). EXR compression results in about half the size.
How much more image latitude, practically speaking, will you retain, while working with 16 bit Lin vs 10 bit Log? Even working with film latitude and DPX 10 bit Log I had never had any issues of running out of dynamic range of the image. Is increased amount of data really worth it, financially speaking?
|« Previous Thread | Next Thread »|