hey guys, so I was rendering out a scene for the last 12 or so hours, it was almost complete, then something went wrong and blender crashes.
I rendered them out as png sequence, so I still have the frames, but how do I now get the vector pass pack ? I don’t know how to import the images back, my only option seems to be importing an image sequence, but there is no vector output socket on it. It only appears on the render layers node
help?
Compositor, use a file input node. Those png files won’t have any vector information so you will have to render that out again.
aaaaaaaaaaahhhh 
12hrs wasted.
what file type holds on to vector data, .exr?
You only need render the vector pass, nothing else. Multilayer OpenExr holds vector data
thanks for your help, but i’m confused, how will the two be combined in order to create blur if do them separately? I thought the info was contained within one image file.
also i’m trying to render just the vector pass, but seemingly its looking like would take the same amount of time to calculate before and after frames. it still renders the image, unless you mean use exr for just the vector pass?
i’ll go with multilayer exr in the future then, I didn’t think i’d get a crash.
If you’re using cycles, switch over to blender render. Then in the Layers tab just enable the Vector pass and render out to an image sequence (Since you don’t need them, also turning off ray tracing, shadows etc in the render settings will also speed up this vector pass rendering). Then you can combine with the original image sequence in the compositor with this vector pass. Wherever you need the vector value just connect to this vector pass image sequence
The Mike Pan BMW test scene renderers just the vector pass in <2secs a frame for me so it is a very fast render
yeh I’m using cycles, i’ll try what you have suggested. thanks for the help
The MultiLayer OpenEXR file format is one that was designed (by Industrial Light & Magic, with a little subsequent help from the Blender Foundation(!)) specifically as “a data-file format for 3D rendering pipelines.”
OpenEXR stores “collections of data as floating-point numbers.” Then, MultiLayer enhances that idea to store “multiple named layers of” data in a single file. (The Blender Foundation originally came up with the “MultiLayer” enhancement, and made it freely available to the graphics community.)
Anyway … “PNG” is just a graphics format. The total extent of its knowledge … of its intended scope … is to store a 4-tuple for each point in an (X,Y) grid: (R, G, B, A), where the value of all four variables is in the clipped-to range [0.0…1.0]. Which is “all well-and-good for PNG, but not sufficient for your purposes (or ILM’s).”
Hence: MultiLayer OpenEXR. All of the RenderLayers, and all of the selected channels of data within each layer, will be present in that file … and any floating-point value (“blacker than black,” “whiter than white”) is equally welcome. The files, of course, will be big, but who cares. (Lossless compression is applied, but “file size” is not a primary concern. No data will be lost or compromised.) The data doesn’t pretend to be “an image.” It is: a numeric data-set. Emitted by a computer, frame by frame by frame, meant for a computer to consume. “Perfect.”
You should structure every stage of your processing pipeline around this format. (In the very final, “answer print” stage, you might just use OpenEXR … but I wouldn’t.) “The finished movie,” therefore, will consist of yet-another directory of these files.
Then, to generate the “deliverables,” in (say) QuickTime or .MOV or what-have-you, you use another set of blend-files whose sole purpose is to read “the finished movie” as input, and to generate “a deliverable” (in whatever format is requested) as output. Kindly notice that this is the only point in time at which anything has been “compressed” or “mapped” or “thrown away,” and that by this point in the pipeline it no longer matters.
Missed this post, sorry. Thanks for the knowledge, when using passes I will use nothing but exr in the future.