Trouble processing files > x3000px on macOS

[ablwr]

Perhaps its the management of memory, but my average-processing-power MacBook seems to not be able to handle large-resolution files. If I convert the file and reduce the width (or likely the issue is file size), ./lenticular will run.

The program does not stop until later, and the error message I get is:

RECONSTRUCTING COLORS...
Computing image dimensions for color reconstructed frame...
rgbFrame.width = -1, rgbFrame.height = 1
ERROR: Won't allocate memory because 'width < 1 || heigth < 1'.

[retokromer]

Hmm… I did process a 4K scan on a MacBook (Retina, 12-inch, Early 2015), that‘s for sure. Yet I cannot check the details now, while in Bangkok, because I don’t have the files with me. Sorry!

[ablwr]

As you noted, it could be the file itself, since it is not returning expected results, as you saw.

[retokromer]

Possibly the film has been shot without the RGB filter on the camera lenses.

[cluelessrgb]

@retokromer I am having the same issue @ablwr is having, except my rgbFrame.width is:
= -532094929, rgbFrame.height = -2613725

I am clueless about coding, but I've made it this far...

How do I fix it?

Thanks!

[retokromer]

Thank you, @cluelessrgb, for your interest!

We at AV Preservation by reto.ch use another approach in production.

I had the intention to continue working also on this approach here on GitHub, because it’s an interesting one. The code has been further developed by the University of Zurich under the direction of Barbara Flückiger and with the collaboration of Joakim Reuteler, the original developer. Before this started, I proposed to David Pfluger to join the forces, but that has been refused and therefore I gave up. In fact, it makes no sense that I sacrify my spare time when the same work is done in parallel by a public founded Swiss university… I suggest you contact Barbara Flückiger by email at baflueckiger@gmail.com or on twitter under @flueko.

[cluelessrgb]

Thank you, @cluelessrgb, for your interest!

We at AV Preservation by reto.ch use another approach in production.

I had the intention to continue working also on this approach here on GitHub, because it’s an interesting one. The code has been further developed by the University of Zurich under the direction of Barbara Flückiger and with the collaboration of Joakim Reuteler, the original developer. Before this started, I proposed to David Pfluger to join the forces, but that has been refused and therefore I gave up. In fact, it makes no sense that I sacrify my spare time when the same work is done in parallel by a public founded Swiss university… I suggest you contact Barbara Flückiger by email at baflueckiger@gmail.com or on twitter under @flueko.

I'll email them. Thank you so much!

[torbjornbp]

As you noted, it could be the file itself, since it is not returning expected results, as you saw.

Did you ever determine whether the source material had any colours, or the script was giving you trouble? I've just gotten the code to run, but I'm getting similar results.

[cluelessrgb]

So sorry, just now seeing this! Apparently the scan was done with the emulsion facing the wrong way… I still have not received the corrected scan, as the person who was doing it stopped responding until recently with no explanation. I’ll let you know and keep you informed!

…

On Oct 28, 2022, at 5:16 AM, Torbjørn Pedersen ***@***.***> wrote: As you noted, it could be the file itself, since it is not returning expected results, as you saw. <https://user-images.githubusercontent.com/3260492/38460128-306639c0-3a82-11e8-9769-4e4b77893e52.jpg> Did you ever determine whether the source material had any colours, or the script was giving you trouble? I've just gotten the code to run, but I'm getting similar results. — Reply to this email directly, view it on GitHub <#5 (comment)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/AQRJ6HCYX2MRXRNMLMKPPQ3WFOKX5ANCNFSM4EZJZE2A>. You are receiving this because you were mentioned.

[torbjornbp]

I'd reckon it would work as long as the lenticules are identifiable. I got it to run in the end, but only get anything with natural colours with the -relaxRaster parameter turned on. Still struggled with loads of artefacts (thick coloured lines and distorted colours for most frames)

[retokromer]

Apparently the scan was done with the emulsion facing the wrong way…

This is irrelevant. It works either way.