Stefan IhringerInefficiencies in VFX: once again a studio sent us thousands of uncompressed EXR files for lens grids and frame leaders. They just let the camera run for 1 minute on 24fps and repeated that for a dozen lenses. More than 1 Terrabyte. But we only need a single frame for each lens and it could have been a 200 kByte JPEG. 😳
Don‘t want to know what the client pays IBM for their Aspera cloud storage and egress fees.
Oh well, our upcoming client doesn't provide cc files for each shot. Instead, we need to extract the grading values from an EDL file. Fortunately I've already written a script to do that a few years ago for another show.
The time to write a script might be more than what it takes to do the task manually (here it would be copying values from a text file to an xml file). But it pays off if you have to repeat the task. Even if that is 7+ years later.
Honestly, there is no better format to show other departments a reference of what you're seeing on the screen right now. If only Jpeg could store timecode metadata...
I don't remember exactly but I think the tracking was part Nuke 3D tracking plus additional stabilisation using a planar tracker for each bag.
All the cuts were from photos we took using a blue plastic bag and I blew some bits of hamster hay into the air for additional elements.
I remember that I hand-animated a corner pin for every cut to make it stick perfectly to the deforming sacks. Frame by frame, hundreds of them.
It’s only „pixel fucking“ if your client feedback is about chromatic aberration on pixel-sized specs of fairy dust. Everything else is just client feedback.
#NukeStudio should rename „sequences“ to „timelines“. Sequences in #VFX are either image file sequences or a collection of shots that have been arranged in sequence. The confusion for editors and artists alike is real.
There’s even a „sequences default framerate“ option in Nuke Studio‘s settings. It does NOT set the default for sequences (timelines) but for sequences (a bunch of enumerated image files).
Buying another company‘s #VFXpipeline like you‘d buy a house:
„Why are there two faucets in the bedroom?“
„Legacy support for the 2019 version of our animation software“
„Ok, and why is the TV upside down?“
„Initially our whole CG team preferred to work upside down but they‘ve all left by now and the TV is the only part of the living room we haven’t refactored yet“
There was this show where we had to exactly match font and position of burn-ins to a given template. Fair enough! Client doesn't want text jumping around if multiple vendors are working on a show.
The template had the name of one of the world's top VFX shops on it! But while trying to match it I noticed that every text field (frame counter, shot name, date) was using a different kerning. I don't know if that was the client's first test to see who's a proper pixel fudger...
Today I started to build the #VFXpipeline for a show that needs to return quicktimes in 2 different resolutions, each using a different codec and each using a different crop of the full resolution EXRs.