[Gimp-user] tips on working with gigantic files?



Hi -- I'm trying to do some simple work on a 0.5GP image in GIMP 2.9.5 (on Linux). (Image is ~23k by 22k pixels, 16bit RGBA).

I can open it fine, and it currently has two partial-canvas layers that are cross faded into each other (I'm manually stitching two slices of a big hugin panorama.)

The RAM is just about maxed -- I have 16GB, and it's enough to do some simple things to the image but I can't do much else.

I need to flatten the image into one layer so I can have a hope of doing further processing on it, but just having the image open puts it too close to the RAM limit to make that possible: I can't flatten the image, I can't export as PNG, etc. Whatever I try, the RAM soon maxes out and the system grinds to a halt, necessitating a hard power cycle.

I tried using 'convert' to convert the .xcf to PNG; no luck, as it doesn't work with 2.9.X XCF files. Maybe there is another route along these lines? (For some reason GIMP won't let me save as an older-version XCF, saying that the file is using incompatible features; not sure what: it's just two layers, one with a layer mask. The layers aren't full canvas size, maybe that's it?)

Any tricks I might be able to pull? Splitting out individual channels? Managing or virtualizing memory somehow? (I notice that my swap is never used -- not sure if it's misconfigured or if it's just not possible to use it in this case for whatever reason...) I don't care if the export procedure is slow, I just need to do it once to get the image down to a manageable size in RAM... maybe I could whip up a VM and give it tons of disk-backed virtual memory?

I tried all this on Windows, too (using a Partha GIMP) -- it actually exported something, but it was mostly black with bits of corrupt tiles here and there. Then I tried again with a larger tile cache and it crashed the OS and corrupted my bios somehow... yikes.

Unfortunately I can't afford more RAM.

Thanks for any ideas!

-c




[Date Prev][Date Next]   [Thread Prev][Thread Next]   [Thread Index] [Date Index] [Author Index]