What file format do you render to?


I’ve been using targa sequences for animations for clients in SD and HD in the past. I have done little post work and use premiere pro to encode.

I’m beginning to move some of my workflow to photoreal vray rendering from scanline and also upping the size to 4K renders.

From the training I’m getting, it looks like OpenEXR is the way to go now for rendering image sequences. I’m guessing because the file size is less than targas, it is still lossless, and it contains some exposure control that I wasn’t sure was in the targas, although it may have been.

I’m not sure premiere pro cs6 can import OpenEXR, but after effects cs6 can, and since photoreal work requires post work, i.e. color correction, exposure adjustments, et. al, I was wondering what you all use to render your image sequences to and why.


EXR is definitely the norm among the pros, but I’m not a pro so I only get to use JPG and PNG.


Yeah, I’m learning post work and images really come to life with adjustments like curve and exposure control, among others, where having that info in the exr makes it crucial. I’m not that good at it yet and I think the advantage over other formats makes that extra bit of difference. I’m guessing that’s why ILM made it, along with saved file space.


For the best flexibility in post use EXR, if you’re using max 2014+ there is an ‘Automatic Gamma’ setting that should be on to ensure the gamma gets set to 1.0 for EXR files. You can save your EXRs as 16bi half-float rather than 32bit to save file size and speed up comp workflow, you rarely need the full 32bit. If doing animation compositing in Fusion or Nuke you can set the save mode to region and make the file sizes even smaller.

We don’t include render elements in the EXRs as it makes the file-sizes too big so we save to separate files.

Hope that helps!


I moved from jpeg (when storage was still a problem and comping very basic) to 16 bit Tifs to 48 bit PNG.
I haven´t found a single time where I actually needed the full 32 bit and pngs just gave me the best performance while comping in After Effect.
I´d love to just output to EXRs with all my render elements, but the workflow for importing and setting them up in After Effects is a pain in the ass.

I´m a little confused now though: I thought there was no difference in bit depth if I output 48bit Pngs compared to 16 bit EXRs, but now I read that PNGS only offer up to 8 bit per channel (+Alpha), while EXRs offer 16 bit per channel…Is this also true for 16 bit EXRs or only for 32 bit EXRs and how do 16 bit EXRs compare to 48 bit PNGs, when it comes to filesize and performance?


Here is a good read on image formats:


I think I’ve found/learned when you want additional exposure (intensity) control when compositing, starting with 32 bit works best. Especially when down converting at any stage in your comp flow…for example lets say I use exr half float, which photoshop reads in at 32 bit, so I think these are like a 32 bit image but without the full float file size. So I’m starting wtih 32 bit, I have very high range exposure control for comp in after effects and premiere pro, then lets say I want to send to speed grade for color grading. AE and PPRO CS6 keeps things at exr 32 bit-ish, going to speedgrade I have to render to dbx files which are 16 bit, so I’m still good at 16 bit. I color grade then output to an intermediate avi at 10 or 8 bit, then final h264 encode to 8 bit avc…or something like that. I’m still experimenting if I can encode straight to h264 out of speedgrade, but right now I use handbrake for final thus the reason for the intermediate.

Your probably fine with 16 bit starter files, but why not get the advantage of half float exr if the file size is comparable, especially if you wanted to adjust exposure at the start. But I haven’t tested png and exr half float files size comparability.

This is my limited knowledge…it may not be fully accurate.


In VRAY, I generally render animations in a JPEG sequence and then compile them in Sony Vegas. But yeah, JPEG is what I use.


Do note that different formats use different ways of describing their bit depths. An 8-bit JPG means it has 8 bits of color per channel, and a 48-bit PNG actually means 48 bits in TOTAL; 16 bits per RGB channel. So 48-bit PNG I believe is the same as 16-bit EXR in this regard. I might be totally off for all I know though, so someone please confirm/defirm this.


except that png uses integer colors and exr uses float values.

in short: png clamps colors/exposure, exr is unclamped (much better)


Ah yes, that I actually knew but had forgotten :slight_smile: Thanks.


except that png uses integer colors and exr uses float values. in short: png clamps colors/exposure, exr is unclamped (much better)

Ah, great, thanks for clearing that up, thats whats always caused my confusion about bit depth…

I´ll just try it out on our next project to see how EXR compares tp PNG speed wise…We´re usually working directly from out servers, because setting up proxies to speed up compositing just is too much setup work, so file reading speed is very essential while doing comps for us and Png seemed to do pretty well in that regard.


You’re welcome.

Keep in mind there are different exr compression options available in max.
I can’t remember right now which one gives best speed/performance when compositing but if you do a quick google I think its fairly easy to find (might also depend on which comp software you use).
Choosing the most optimal compression for speed when comping might not be the one that gives the smallest file-size though…


I’m always rendering to TIFF. It has 300 dpi (or more if you want) and alpha channel, all what I need in “case of emergency” :slight_smile:


Choosing the most optimal compression for speed when comping might not be the one that gives the smallest file-size though…

Yeah…I think we´ll stickt to PNG for now. EXR in After Effects just performs terrible…simple black and white EXR image sequence in 1080p renders in AE 10 times slower than PNG. I also tried different compressions (zip,RLE, no compression) and the differences were´nt really notable.
And I´m not even using multi channel exrs.

Also the only times I actually want to have more control usually is when adjusting the Zbuffer and then I´d either have to work with 32 bit files for all channels or not use the Vray frame buffer and set each channel up separately.

So thanks for the input, but as long as we´ll be using After Effects and they don´t fix their handling of EXRs, we´ll have to stick to other file formats…