Intruder said:

towne32 said:

FrankT said:

I wonder how superresolution can treat motion smearing artifacts. Thinking about it, they'd probably stick out.

I cleaned up the speeder shot a bit, so hopefully that one (the worst artifacts, IMO) will not look quite as terrible, at least.

Did I miss anything? I have read reference about your version, but cannot find the source to it.

It's not available. Just something I was working on for my own use while Dre was uploading his tests here. It could have used another pass, especially in the final 30 minutes of the film. But Dre needed to start encoding and wanted to use it. I hope that it doesn't disappoint, as he has spent forever in render time. I admit I became much more nervous after learning how long it would actually take to encode, as he can't simply whip out a new version... :). The sample posted today uses the cleanup, afaik.

edit: I guess I should say what it is. Just manual dirt, scratch, and dust removal. Nothing automated. Glue removed only when an adjacent frame has aligned picture (i.e., nothing intricately rebuilt). 3 passes on the first 30mins. 2 on the next hour, 1 minimal pass (mostly due to the complexity/speed of the death star run) on the final 30 minutes. Knowing what I know now, it would have been more thorough and much more efficient in regards to time spent on it. Oh well, live and learn. :)