Problems with Gigapixel AI and after the update, still happening. Whenever I start it, a get a series of error messages: “The procedure entrypoint cLEnqueueFillBuffer could not be located in the dynamic link library OpenCL.DLL”
Hitting the X to close it several times eventually it stops appearing.
Windows 7 Pro, INTEL i7 and 32 gigs RAM, NVIDIA GT Force GT640 video
Have posted request on Topaz Help a few times but do not receive any responses…
UPDATE: I got updated drivers for the GEForce GT640 and that seems to have fixed the problem…
When we can see update in ai models for gigapixel currently there is application cloud based ai image enlarger produce better result.
But I like gigapixel I don’t like cloud based apps.
And is it possible that we can train our own model ? Like if we add images it will update the current models used and show that is there is any improvement or not ?
Another idea is like if we feed a similar high quality image like of same type of scene or same person as a reference image to improve the quality of enlarge ?
You gave me more speed and productivity with one hand, then took away a large portion of that with the other. I like the way Gigapixel is running. I like the new UI. But I used to be able to set “default” settings for processing and as long as I didn’t quit Gigapixel, I could continue to use those settings on other images. Now it always resets back to your default, not mine. Why?
I’ve always loved Gigapixel AI for how it can transform images, however the latest version 4.5 has clunky functionality which makes it overall a lot slower to work with (unless you are using it for batch processing only). It doesn’t keep any of the settings from one image to the next, unlike 4.4.5. For me, it’s a disappointment.
Oops, sorry. Should have included that; I’m on a Mac with Mojave 10.14.6.
Uninstall procedures vary; sometimes there’s an “Uninstall” app along with the target program. There is one in “Topaz Labs LLC”, there doesn’t appear to be one in “Topaz AI Gigapixel.”
In this case I would normally just move “Topaz AI Gigapixel” to the Trash. That would remove the main files, but the install process for the new version asks where to put secondary files, I think in "User->Library->Application Support->Etc. So there seems to be more stuff that’s not in “Applications.”
Obviously this update was rushed out without much thought given to user workflow. Hoping that future 4.5.1 or 4.5.2 will fix some of the most glaring functionality flaws. I am not going to mention which since the previous comments have already covered it.
I think you should use Jpeg to RAW first to get the maximum color depth and remove jpeg artifacts. Gigapixel will often remove artifacts but not guaranteed.
OS X version is unusably buggy - freezes on generating preview, if you turn off autogenerate preview and move the preview area around, the UI glitches out and becomes unusable. If you try to generate images it just freezes. Wish I hadn’t deleted the prior version before trying this…
EDIT: This appears to be due to forcing the app to use the eGPU, (radeon rx-570) which previously worked perfectly. When I unchecked this it went back to using my internal GPU (geforce GT750m), although it actually does function with the internal GPU.
Wait, we can’t no more drag and drop a folder containing images ? “File type not supported”
Also not keeping previous settings and going back default each time loading a new picture is annoying.
Same when saving files not keeping the previous settings.
Hi! Offline installer is available as download links in this thread. (For future releases, we’ll always have the offline installer available in the forum update thread.)
Thanks for the report! I can’t duplicate this behavior, unfortunately - it appears to start processing correctly for me. Would it be possible to send a screenshot or screen recording of this?
For permissions reasons, drag/drop won’t currently work in Windows if you open Gigapixel from the installer checkbox. If you open it normally, though, it should work correctly. I’ll add this as a known issue, we’ll look at fixing it for a future update.
Can you send me a link to the image over PM?
Good point. We’ll explore adding this in an update.
Gigapixel currently doesn’t support LR / PS integration (unlike DeNoise / Sharpen).
Looks like there’s an issue where it doesn’t show up sometimes - will fix in update soon!
Thanks for all the feedback so far everyone, will be working on updates to address some of the issues you’re seeing.
In theory, yes, but in practice, no. Because JPEG to RAW adds its own artifacts (such as chromatic aberration and sometimes over sharpening, which Gigapixel amplifies. I find the reverse works better to clean up further the Gigapixel result.
AI Gigapixel is way far superior compared to Jpeg to Raw when it comes to remove jpeg artifacts, especially if the jpeg image has chroma subsampling. And yes Jpeg2Raw adds too many artifacts and you will end up with a worse result if you use it before AI Gigapixel. In fact it’s the same if you try to use any other else AI program like Denoise or Sharpen before. Ai Gigapixel will amplify any artifacts or approximations done. The best workflow imho if you want to use the others AI programs with an enlarged image is to use AI Gigapixel first with zero denoise and sharpen. Talking about that, the new version seems not to have the possibility to bypass any postprocessing. The minimum values for denoise and sharpen are set to 1. I know this is almost the same as zero but why we can’t go zero by now ?
Thank you for the information. I didn’t know this about color depth. I used GigaPixel first, then JPEG to RAW on some Saturday, but didn’t see an advantage I’ve just GigaPixel. In fact it was worse afterwards. As you suggest, I’ll try it the other way around.
Other than supporting a company who’s products I like, I was wondering if purchasing JPEG to RAW was worth the purchase. It was the only product that I didn’t already own. BTW, GigaPixel did a fantastic job on the landscapes by itself. Other than adding a little blur to skies, it made a nice printable file from a 175K file.
