Distortion workflow


Problems with mocha ofx for fusion again.

It starts with working on anamorphic footage.
Mocha does not correctly read the PAR settings from fusion so you have to do a counter squeeze in fusion with a resize node to get image display correct in mocha.

Now I am trying to get a working lens distortion workflow and it’s making me nuts.

Mocha very temperamental in lens module on anamorphic. I have to not do the above counter squish but rather open clip squeezed in mocha, add a new clip and change PAR to 2. Then everything views correctly in that session. Next time I open have to redo this step.
Like this the lens module works better.
I calibrate the lens and export stmaps.
Great but now I’m fusion I’m really struggling to apply them. I am using stMapper node. Undistort works fine as expected but when I then redistoet I loose the corners of the image and there is significant quality loss.

I spoke with the creator of stMapper. He says he ran test. Works fine with stMaps from other software but mocha distort stmap is double distorting- not sure why.

This is terrible. I rent 5 licenses. I have a lot of anamorphic shots to do. If you can help please do.


Hello, could you please send me one of your footages that you are working on, and STmaps that you’ve rendered? That will help dive into a problem.

You can send it privately in DM if you don’t want in public.

We’re awaiting a response from the BMD team on this part. We are being sent the wrong value by Fusion.
Oddly enough if you REVERSE the x/y PAR input in the Fusion loader you get the right result in Mocha, but that doesn’t help your output.

In terms of the calibration, as @elizabeth.postol says, are you able to send me your setup and files?

What is your goal with the anamorphic footage? Do you want to undistort the source then re-distort, or are you trying to apply distorted comps to the original source?

Thank you for these responses. I will make and send a sample tomorrow.

Look forward to BMD response on PAR.

Either should be possible if the workflow is correct no?
I would be creating added elements in the undistorted layout and then redistorting and comping back over the original source.
This may involve cleanup/paint on undistorted plate and or added elements.

Ok, i’m just checking because you can avoid an extra step of the workflow if you just apply the distorted stmap or the Mocha lens distort render to the painted/inserted elements rather than undistorting then re-distorting.

Having said that, we are reproducing an anomaly with the re-distort workflow when using stMapper in Fusion. I’m going to file a bug and see if we can get to the bottom of this as there’s definitely a mismatch going on here.


any progress on this issue?

@elizabeth.postol I am going to private DM you a frame now with my test comp to show you my issue - see if its the same you are getting.

ANY CHANCE we could solve this by next week please? I am suffering here :slight_smile:

I don’t want to leave you in the dark, so i’ll give you an update.
So yes, it looks like the STMap is currently clamping to the bounds of the original image. The main issue is the extreme edges of the plate where you lose information at the corners after it is re-distorted.

We may have a solution to the problem which is currently in development. We’re just about to release maintenance update v9.5.5 so we’ll have to look at fixing the lens distortion issue right after that one is out.

While we fix this particular problem you may need to find an alternate way to do this. Unless you’re not painting into the edges, I’d suggest using the built-in lens models in Fusion and applying the Mocha to the undistorted plate instead.

Hello thank you for the update.
It’s a rather gaping issue to not be working on a tracking software, glad you are onto it now. It’s a pity my older copy won’t work unless I pay to update it.

I did find a way to make it work fully, a workaround I sent to Elizabeth to show her. Perhaps you can look at it