JPG file format dataset.

[dhruvi18022004]

Hello, my dataset is in jpg format and i want to use this same algorithm for my project, there is some issue regarding jpg files. This algorithm only allow dng files. beacuse it gives me error: atleast 1 raw .dng file in your burst folder.Is there any solution for this?

[Jamy-L]

Hi, the algorithm only really works with the assumption that the input frames are raw, and have not yet been through the whole image pipeline that produces a JPEG. If you only have the JPEGs available this algorithm will not work at all for diverse theoric reasons, and I would recommend to use a different approach.

As a curiosity, may I asked in what context does one work with JPEG bursts?

[dhruvi18022004]

I created my own dataset by mobile camera, which was already in jpg format, now I wanted to use this algorithm on my own data. so that.
1 question is that is this algo suitable for png format or any other file format or only for dng format dataset?

[dhruvi18022004]

One more question I have is, should I convert jpg format to dng format and then place into the test _burst folder? will it work?

[Jamy-L]

Thank you for the details. In your case, your phone has already done a lot of processing to transform the raw image into a jpeg. This transform is not revertible because it modified the pixels and removed some essential metadata that would have been present in a real raw .dng.

While you could in theory transform a jpeg to dng, I assume the code will not work because some crucial metadata will be missing. And even if it could run, the result would be pretty bad because your .dng would not be a real raw image.

I advise you to try a different approach

[dhruvi18022004]

okay, Thank you.

[dhruvi18022004]

one more I have a question is that, can you tell me where you created function for take multiple raw input raw images and then generated 1 super resolution image, in which python file it is available?