I’ve noticed file from A1 and A7r4. Render the image worst than original. Especially with the animals hair. It’s almost like it shifts the image and blurs it. But when processing with Topaz sharpen zero issues. Anyone else see this issue ?
Topaz DeNoise and High mega pixel issue
- Thread starter Jerryjaws
- Start date
If you would like to post, you'll need to register. Note that if you have a BCG store account, you'll need a new, separate account here (we keep the two sites separate for security purposes).
KirstyM
Well-known member
Are you using the "auto" function or adjusting manually? I know that Matt Kloskowski shoots an A1 and is thrilled with both denoise and sharpen for his BIF shots....
I'm a Nikon shooter (many less megapixels!!) and have actually found the reverse to you - denoise produces a much pleasanter sharpening result than sharpen, which seems to introduce artefacts however gentle I am with the sliders! Not much help, I know.... sorry!
I use the auto function and manual backed all the way down. And still have issues with DeNoise. Topaz sharpen zero issues.
I find both Denoise and Sharpen have trouble with fur if pushed too far. Seldom can you push the sharpening above 40 without the fur starting to get artifacts.
The other question is are you importing Jpeg or Raw? Also are you exporting the image from Lightroom or loading it straight into Topaz?
Raw from LR to topaz.
One thing I do is make sure the sharpening and noise reduction sliders in the details tab of lightroom are zeroed out. I've found that if there are default values it does funny things when I move it to Topaz
That’s the first thing I do. Theres a issue with high MP sensors and there program. My A9 works prefect. Soon as I load A1 or A7r4 into program. Issues
Interesting. I process Z7 images through it all the time.
Backdoor Arts
Well-known member
I'm using it with an R5 and not seeing any issues. I avoid auto-anything at all costs and have developed a feel for where to set sliders for a particular image. I also avoid the original detail slider and instead invoke it from Photoshop as a new layer and if details are obliterated in Denoise then I paint them back using a layer mask. No tool is perfect but Denoise has so simplified my workflow.
lwan
Well-known member
The way topaz handles very fine picture features like fur and feathers is the reason I moved to DxO instead. Topaz fattens them to a point it's not natural-looking anymore. Haven't tried the latest releases but I'm happy with DxO's deep prime stellar denoising and sharpening.
Blue tit photo
Well-known member
It often selects the raw mode. Use AI Clear because the prior is terrible. Though as everybody said dxo is better
My Topaz DeNoise works really well.
Are you comparing images inside the software? Sometime the preview is not 100% accurate but the output tends to deliver.
Are you comparing images inside the software? Sometime the preview is not 100% accurate but the output tends to deliver.
lwan
Well-known member
Sure, this is a bit of an extreme example as it was shot at ISO 2000 through a single pane glass, and the only way I would display this file would be downsampled but it shows that typical problems that occur with topaz denoise for me, since AI clear to now (v3.3.1 I think). Both raws processed as they were, exported to DNG adobe color profile applied in ACR (that's another thing, Topaz does not retain any camera color profile information while DxO does a good job at making the raw look as it was displayed in-camera). Top row is 100% crop of the straight raw to jpeg conversion with nothing edited at all, second row denoised 100% crop (default setting in both software), third row is a blow up. Pay attention (enlarge bigger than 100% if needed....):
- out of focus noise is nice and regular in DxO / blotchy and artifacted in Topaz
- fine noisy feathers are reconstructed quite well in DxO (for what the input is...) / fattened and smudged in Topaz. It's like topaz can't really tell "lines" and wave its way into something that does not exist at all, while DxO doesn't do that. Probably the advantage of denoising at the demosaicing stage.
The fattening of small details becomes more and more unnatural as you downsample or print, imho. You can work your way around it, but I'd rather save myself more computer time.
- out of focus noise is nice and regular in DxO / blotchy and artifacted in Topaz
- fine noisy feathers are reconstructed quite well in DxO (for what the input is...) / fattened and smudged in Topaz. It's like topaz can't really tell "lines" and wave its way into something that does not exist at all, while DxO doesn't do that. Probably the advantage of denoising at the demosaicing stage.
The fattening of small details becomes more and more unnatural as you downsample or print, imho. You can work your way around it, but I'd rather save myself more computer time.
You can only see EXIF info for this image if you are logged in.
Blue tit photo
Well-known member
dxo is much better and it will always be
lwan
Well-known member
I would be glad if that statement turned out to be false, that'd mean something else got even better
But yeah, I think DxO's approach is better and is probably possible because of their installations and equipment.
Turberticus
Well-known member
I have similar issues when it comes to fine lines like feather detail 'fattening' especially in AI Clear mode with Topaz Denoise. I will go back to the standard Denoise AI if that happens and mess around with the sliders until I can get a natural look. I'll have to try DxO.
lwan
Well-known member
Personally although AI clear was a bit of a revolution in denoising in 2018, the regular denoise that followed was much better. It's just that for some time (I think until version 2.5?) denoise needed the raws to be converted to jpeg or tiff with color noise removed (AI clear was good at denoising color noise).