Hello, I have question regarding slowdown rendering previews and exports using Gigapixel and Upscale in Photo AI. Currently, my setup is a 2019 Macbook Pro, Intel Processor, 2.3 GHZ, 8 Cores, 64 Gigs of Ram, and AMD Radeon Pro 5500M 8 Gigs of VRAM. Let’s say I want to resize an image that is 5248 × 3499 pixels, roughly 17 x 10 inches, to 24 x 16 inches, or 7200 x 4800 pixels. This is a 1.4x upscale. I have Gigapixel to run off the GPU. The model AI Gigapixel chooses is High Fidelity for this type of image. For this image to render to my specifications, it can take up to an hour to export. Rendering the preview can take 30 minutes or so. Now, the odd thing, is if the image is lower in resolution, and I resize it using Low Resolution mode, and 2X resolution, the processing time is much faster/acceptable.
Is this normal behavior, for Gigapixel, in terms of preview and export times, given my computer specs? While a upgrade on my current computer is in order, do rendering times for new Macbook Pros with the M4 chipsets perform better with Gigapixel? Gigapixel used to perform well for my machine in the past, but the new models/software cause my machine to run at a snails pace.
I have a similar issue. I’m using the following macbook (2019 MacBook Pro / 2,3 GHz 8-Core Intel Core i9 / AMD Radeon Pro 5500M 4 GB Intel UHD Graphics 630 1536 MB / 16 GB 2667 MHz DDR4).
Right now it’s almost impossible to export a simple image, even if I render it at 1x (standard model). I remember a year ago everything went pretty fast. I don’t know if it’s because the updates (using Gigapixel 8) but there is no way I can work with this. For 20 images (around 3000x4000 pixels / 40-70mb) it takes literally multiple hours (5+). Really curious about a solution!
Yikes.I’ve used Topaz Plug-ins for years, and have had satisfactory results. But these last two years or so, have seen a degradation in terms of speed. Would like to know as well about solutions, or if newer Macs with the M4 chipsets perform better.
They’ve built a system to buy credits for an online render service… so what do you expect… In my opinion this is outrageous. As soon as there is a comparable AI software that can be used locally, and uses all the potential of the GPU, I’ll gonna switch…
I think you are correct. They seem to be pushing that model, rather than local processing. I was watching a YouTube reviewer (Todd Dominey), who resized an image using an M2 Ultra Mac. Compared Standard Model vs Recovery. The Standard Model took a minute or so, but the Recovery Model took over 15 minutes (1024 px image upsized 3 times). It took around a minute if he uploaded it to the cloud.
He then did a high quality photo, and upsized it to 60 inches by 40 inches. Did a test between Standard and Recovery. Standard took a few minutes to render. Recovery Model had to run overnight. The file size was too big to run on the cloud. He found that the standard model and recovery were undistinguishable from another.
So, essentially, what I am finding, is the newer hardware helps. But the newer models that use generative functions do not appear to be optimized for local processing.
Just a quick update. I bought a new computer, Apple with M4 Max chip, and the new hard ware makes a world of difference in speed. My guess, is the software is now optimized for these chips, and the Intel based macs aren’t going to cut it anymore
I also have the feeling, Topaz want us to render in the cloud. I have a Mac Studio M2 Ultra with 192 GB Memory. High Fidelity always ran pretty fast for me. Now with the same setting it takes a lot longer than before. What was about 30 sec now takes 5 mins or more.
@TopazLabs: Take care, you will be loosing your Pro Customers…!
