No 16-bit EXR

[chut11515885]

In After Effects, there is no option to output 16 bit for EXR.  It always default to 32 bit.  Our work does not need 32 bit especially our resolution is huge (14,000 x 14,000) so 32 bit will create a huge file.

Thanks.

[Joost van der Hoeven]

If you feel strong about this feature you can vote for it here: Open-EXR 16-bit HALF-float rendering – Adobe video & audio apps

[fnordware]

AE writes out 16-bit EXRs by default. If you want 32-bit you have to check the "32-bit (not recommended)" box in the OpenEXR options.

Maybe you're confused because the output module is always 32-bit float. EXR files appear to AE as 32-bit float even when the file itself is 16-bit float. This is not the same as AE's 16-bit color, which is integer, not float.

[jaydude-wb]

Not true, even when the checkbox is off, and Ae is in 16bit mode, at render time a warning pops up that 32bit is more than the project setting's 16bit.

And when choosing OpenEXR, all depths but Floating Point are grayed out. So 32bit is the only possible choice.

[fnordware]

After Effects's 16-bit mode is 16-bit integer. EXR's is 16-bit float. That's why EXR only lets you use Trillions (32-bit), because that's the only floating point mode in After Effects.

Unless you have a good reason, you'll want to stick to 16-bit in EXR. Good reasons would be 1) you have pixel values greater than 65504.0 or 2) you're writing data like z-depth that reqires full 32-bit precision. For conventional images, 16-bit is plenty and makes a smaller file.

You can save EXRs from an 8- or 16-bit AE project and AE will do the linear conversion for you. Naturally you won't have the same fidelity as a 32-bit project and won't have any overbrights.

[P.M.B]

Try reading the post again.  The OP's problem is that AE will not let them export 16bit.  It's forcing them to export 32bit.  It was clearly stated in the post that they are aware 32bit is unecessary.  

[fnordware]

Try reading my post again.

You can keep your AE project in 16-bit if you insist, but EXRs are always float, so the output module will only allow float. You'll get a warning about the project depth being less than the output depth, but it will render.

If any of the input footage was EXR, you may be clipping it by using a 16-bit project. If the client wants EXR, there may be other things they're expecting that could get lost in 16-bit, but if a basic conversion is fine, then it'll work.

Actually, with a project that big, AE may run out of memory creating the floating point buffer for the EXR at the output stage. 16-bit will be tough as it is. Might want to render to some other format and then find a way to convert to EXR after.

[P.M.B]

Sorry, guess i didn't understand it/

[jaydude-wb]

Millions = 8bit, Trillions = 16bit, Floating Point = 32bit. EXR can only be set to Floating Point...

[fnordware]

Ack, sorry. I kept saying Trillions was floating point, but it's 16-bit. Floating point is floating point. Earlier posts have been corrected. Thanks.

[stwert]

Just wanted to say, very helpful thread, thank you! I was going down the wrong path trying to figure out why the Half Precision checkbox in Arnold's render settings wasn't working and I was only getting 32-bit EXRs. Typically I use Fusion for compositing (and Redshift for rendering) but I'm teaching a class, so gotta use what the curriculum requires. Also thank you @fnordware for the awesome plugins!