For years I’ve had hopes of making Topaz Studio my go-to, primary tool for RAW development and image processing. It continues to be used in only special circumstances mainly because of one (maybe not just one) thing: it’s slow. Especially with the AI plugins.
I get that there’s a lot going on under the hood, but the thing is constantly “Generating Preview”. Zoom in - redraw. Zoom out - redraw. Move your loupe - redraw. This is all with no changes being made to the adjustment.
Surely once an adjustment preview has been generated for an image, the entire image preview could be cached somewhere to that simply changing your zoom or position in the image doesn’t require reprocessing the preview…
1 Like
Hate to say it, but slow is always relative. For example, Dynamic-Auto-Painter is dog-slow (non-GPU) compared to Impression (GPU). As you stated, the nature of the beast is billions of calculations and that takes time.
For HW on the low perf side, it is almost always preferable to have Auto-Update turned OFF with the AI apps. And update (redraw) manually when desired ie with the Update button.
Otherwise, there is not really any comfortable solution but a well performing video (GPU) graphics card. Cards are not that expensive (~$125 can easily get you there on desktop).
For comparison purposes, you might want to utilize the Neat Image video benchmark (free) so as to compare apples to apples with quantified numbers. It’s not possible to make any educated suggestions unless you post your HW specifications.
1 Like
My system basics are as follows:
-
- Intel i7-8700
-
- 64 GB RAM
-
- Radeon RX 570 w/ 8GB RAM
I don’t know how things operate under the hood. Perhaps they only calculate what is on screen at the resolution displayed so it’s faster as adjustments are being made.
If I zoom in or out, or move the loupe/view window, it’s because I want to see the result at that spot. If it’s not accurate because I have “auto update” off and it needs to be redrawn then the zoom/positioning is moot.
With each movement and mouse click taking about 5 seconds to redraw, it very quickly gets tedious. If the entire image was cached with processing applied, simply moving about the image would be instant. Buy maybe that would mean the actual recalculate after each adjustment would take way longer.
Your HW specs are significantly higher than mine and my perf is very acceptable. But, 5 seconds might be acceptable to me, but not to you. It’s all relative due to the number of calcs being done.
What size image?
What numbers do you get when you run this benchmark?
I have no issue with whatever processing time is required when I make adjustments. It’s when simply moving around the image without making any adjustments takes 5 sec per click that gets tedious. I’m hoping the coders will put some thought into how to remedy this.
The RAW files out of my camera are 6264 x 4180, so that would be maximum resolution.
Here is the result of the benchmark
Neat Video benchmark:
Frame Size: 1920x1080 progressive
Bitdepth: 32 bits per channel
Mix with Original: Disabled
Temporal Filter: Enabled
Quality Mode: Normal
Radius: 2 frames
Dust and Scratches: Disabled
Repeat Rate: 0% of repeated frames
Jitter Filtration: Normal
Spatial Filter: Enabled
Quality Mode: Normal
Frequencies: High, Mid, Low, Very Low
Artifact Removal: Enabled
Edge Smoothing: Disabled
Sharpening: Disabled
Detecting the best combination of performance settings:
running the test data set on up to 12 CPU cores and on up to 1 GPU
CPU Model: Intel(R) Core™ i7-8700 CPU @ 3.20GHz
GPU 1: Radeon RX 570 Series (OpenCL): 8192 MB total (8142 MB currently available), using up to 100%
CPU only (1 core): 3.2 frames/sec
CPU only (2 cores): 6.1 frames/sec
CPU only (3 cores): 4.95 frames/sec
CPU only (4 cores): 6 frames/sec
CPU only (5 cores): 5.01 frames/sec
CPU only (6 cores): 6.34 frames/sec
CPU only (7 cores): 5.5 frames/sec
CPU only (8 cores): 5.5 frames/sec
CPU only (9 cores): 5.74 frames/sec
CPU only (10 cores): 5.77 frames/sec
CPU only (11 cores): 5.8 frames/sec
CPU only (12 cores): 5.81 frames/sec
GPU only (Radeon RX 570 Series): 16.7 frames/sec
CPU (2 cores) and GPU (Radeon RX 570 Series): 13.1 frames/sec
CPU (3 cores) and GPU (Radeon RX 570 Series): 16 frames/sec
CPU (4 cores) and GPU (Radeon RX 570 Series): 16.9 frames/sec
CPU (5 cores) and GPU (Radeon RX 570 Series): 16.6 frames/sec
CPU (6 cores) and GPU (Radeon RX 570 Series): 16.4 frames/sec
CPU (7 cores) and GPU (Radeon RX 570 Series): 15.6 frames/sec
CPU (8 cores) and GPU (Radeon RX 570 Series): 15.1 frames/sec
CPU (9 cores) and GPU (Radeon RX 570 Series): 14.6 frames/sec
CPU (10 cores) and GPU (Radeon RX 570 Series): 14 frames/sec
CPU (11 cores) and GPU (Radeon RX 570 Series): 13.2 frames/sec
CPU (12 cores) and GPU (Radeon RX 570 Series): 10.6 frames/sec
Best combination: CPU (4 cores) and GPU (Radeon RX 570 Series): 16.9 frames/sec
Along these lines…has anyone on a Mac tried using Topaz Studio 2 with the new M1 Macs? I have heard Affinity Photo is a speed demon with it so was wondering if Studio was quicker as well?
On my older Mac Topaz Studio 2 is somewhat slow rending processing but not so much as to make it not worthwhile.
If M1 Macs run the Big Sur MacOS, then TS2 will not currently run on them. I think someone posted as such just a few days ago, but I don’t recall which thread.
For HW on the low perf side, it is almost always preferable to have Auto-Update turned OFF with the AI apps.
I do not seem to have the Auto-Update option on my paid copy of Denoise AI v3.1.2. And it would be nice, for the reason RayC stated. Am I missing something? Is that option hidden somewhere?
In Denoise AI that option is found under File>Preferences.
I think he was asking about Denoise v3.1.x which I haven’t updated to.
You’re right - typo on my part.
