Gigapixel v6.2.2

yes but that might lead to even more confusion when someone owns a certain megapixel camera but by choosen the same megapixels in the application the size will be noticeably different!

The problem with this and why I hate that in PhotoAI is the AI models (as far as Iā€™m aware) are based on scales of 2x, 4x and 6x. Anything inbetween uses the lower AI model. eg 3x would do 2x AI then plain upsampling to reach 3x, which is absolutely not what I want when Iā€™m using an application thatā€™s supposed to be using AI, its just a waste of storage space.

For some reason Face Recovery isnā€™t working for me on 6.2.xx , seems like itā€™s disabled or using an older model as the results I get are poor, had to go back to 6.1

1 Like

Actually, the AI models in all Topaz AI apps are based on only 1x, 2x and 4xā€¦ and 6x is a lanczos upscale from 4x. The transitions between the sizes are at 121% and 240%. In other words, 121% and up uses the 2x model, and 240% and up uses the 4x model.

Below image shows some of the model tile sizes in 6.2.2:

Screen Shot 2022-11-24 at 4.46.42 PM

1 Like

Iā€™m in the same position as you. Face recovery appears to do nothing though it does spend time calculating something when activated for the first time on an image. The same happens in the 6.3 beta versions, but the PhotoAI app face recovery does work.

I have a GTX 1060.

Thatā€™s even weirder given Gigapixel having 0.5x, 2x, 4x, 6x in the UI - would make more sense to stick to AI model presets and I certainly use 1x more than 0.5x and why 6x is generally worthless.

Same happens in video enhance, the HD, 4k, etc resolution sizes donā€™t necessarily correspond to the AI model sizes.

There is at least some logic there, if you are mastering for a resolution standard. Although Iā€™d argue still better off letting your production software do the final scale. I donā€™t see the logic in wasting space on plain upscaling before the final pass (or at all if you arenā€™t targeting a specific resolution for a good reason).

Even if I have chosen to use my GPU to upscale the images, Gigapixel insists on using my CPU cores which are very slow. I double checked it from the activity monitor - my GPU is idle. I didnā€™t have that problem when I had version 5.5.2, in fact the process used to be super fast. Now I have to wait ages for a single photo upscale.

Application & Version: Topaz Gigapixel AI Version 6.2.2
Operating System: macOS 12.6
Graphics Hardware: AMD Radeon VII OpenGL Engine
OpenGL Driver: 4.1 ATI-4.8.101
CPU RAM: 131072 MB
Video RAM: 2047 MB
Preview Limit: 4492 Pixels

1 Like

Gigapixel just doesnā€™t know how to use modern hardware. Itā€™s like it was written in 2003.

2 Likes

Hello, I am using Gigapixel AI to scale up an hdr file of a sky. When exporting the image and then reading it in Nuke, i get a bounding box error. Do you have any ideas why this problem occurs?

bounding box

They found the software in a time capsule from 2003, and thought they could get away with releasing it in 2022. :smile:

1 Like

His 2 GB Vram is like 2007.

Not far off:

Radeon HD 5870, Sep 23, 2009 - Size 2048 MB

I tried out 6.2.2 and found no significant improvement compared to 5.7.1 I used to run before. But I found something I consider a weird bug: If JPG is chosen as output format, an image of, say, 9 Megapixel is saved as a JPG roundabout 1,5 MB in file size, with the respective compression artefacts clearly visible. Saving the same file as PNG will yield a result of 8 MB in file size.

Am I the only one experiencing this issue? If not, when can we expect this to be fixed?

I have a 2021 6700XT with 12GB and get the same result.

Is your driver up to date?

Have a Radeon too and it does work.

Oh its all up to date and runs fine with everything else just Topaz software (Gigapixel and VideoAI) just give worse and worse performance with every version. Been using the software around three years now. I refuse to pay any more subs until things appear to improve.

As Iā€™ve mentioned I donā€™t want to run one video or picture list at 33% CPU/GPU resource. I (and others it seems) want to be able to throw all the computing power I have in one go. Why do one thing over an hour at 33% when I could do it in 20 mins at 100%?

I just downloaded Gigapixel 6.2.2. The big window comes up. I drag .jpg into it and boom.

Model Name: Mac mini
Model Identifier: Macmini8,1
Processor Name: 6-Core Intel Core i7
Processor Speed: 3.2 GHz
Number of Processors: 1
Total Number of Cores: 6
L2 Cache (per Core): 256 KB
L3 Cache: 12 MB
Hyper-Threading Technology: Enabled
Memory: 32 GB

I have the crash report.

You need to check with the Nuke support as to what they are looking for in the app, or try PNG as the export. Nuke is a video editing app.

Iā€™m using it for Hdr images which I think Png doesnā€™t support though?