dxo

[george013]

I purchaised DxO some months ago. I'm still not used to it. I did go to their forum, registered and asked a question, 2 moths ago. Now I did get an answer notification by email but can't login for I forgot the password. Tried the recovery procedure, name is known but not the email!! And I got a notification from them. Tried to register again, succeeded but am still waiting for the authentication mail. This for some one will tell me to go to the DxO forum.

I did notice there're several users of Photolab here. At this moment I got 2 questions.
1) Why do I get a sharper image on my screen when enlarging above 70% and less sharp when at 70% or lower?
2) How can I resize the image and only then apply sharpness to it?

Hoping somebody can tell me.

George

[davidedric]

Hi George, i do use PhotoLab, but not at a level to be able to answer your questions.

However, in Lightroom (different product I know!), what you see on screen is rather complex, depends on where you are in the program and other things. It needs a real technical expert to lay it out.

I wonder if PhotoLab is similarly complex, and so persevering with the forum, or tech support, seems the way to go.

Dave

[Manfred M]

George - whenever you are not looking at an image at full size, the software will upsample and downsample to get it to fit the screen. As these algorithms are all different, there are questions on how accurately they show the image. Most sources I use suggest that all input and in-process sharpening should be done at 100% size to avoid the resampling issues. For output sharpening the suggestion is to view the image at 50%. The rationale is a bit obscure, but as sharpening effectively is local brightening and darkening at the edges, the 50% approach is to compensate for the fact that the human visual system is more tuned to shadow detail than the highlights and this seems to balance out this effect.

While I also use Photolab, I only do so as a raw converter and I do all my sharpening and resizing in Photoshop, so can't help you with these questions either.

[george013]

George - whenever you are not looking at an image at full size, the software will upsample and downsample to get it to fit the screen. As these algorithms are all different, there are questions on how accurately they show the image. Most sources I use suggest that all input and in-process sharpening should be done at 100% size to avoid the resampling issues. For output sharpening the suggestion is to view the image at 50%. The rationale is a bit obscure, but as sharpening effectively is local brightening and darkening at the edges, the 50% approach is to compensate for the fact that the human visual system is more tuned to shadow detail than the highlights and this seems to balance out this effect.

While I also use Photolab, I only do so as a raw converter and I do all my sharpening and resizing in Photoshop, so can't help you with these questions either.

That change from sharp to unsharp happens going from 70% to 80%, independent of the image size. So with the size I post here, longest size 1600, as well with images of the D700,4256x2832, and of the D750, 6016x4016. I haven't seen this with other software.
And I'm really missing the ability to resize the image, see the result and make eventual some corrections. Now resizing is done as part of the writing to the disk process.
If there's no workaround for this I've to live with it as long I use the program.

George

[xpatUSA]

That change from sharp to unsharp happens going from 70% to 80%, independent of the image size. So with the size I post here, longest size 1600, as well with images of the D700,4256x2832, and of the D750, 6016x4016. I haven't seen this with other software.

George, I deleted my previous post which may not apply to your experience.

You said "Why do I get a sharper image on my screen when enlarging above 70% and less sharp when at 70% or lower?"

By "enlarging" did you mean zooming into the review image?

If so, it could be that DxO applies a different re-sampling above a certain zoom level and/or applies sharpening below a certain zoom level. I am no fan of DxO ...

You said "How can I resize the image and only then apply sharpness to it?"

I would open the image and immediately resize it before any other editing. Then when you come to sharpening (normally done last) do it at 100% zoom.

And I'm really missing the ability to resize the image, see the result and eventually some corrections. Now, resizing is done as part of the writing to the disk process.
If there's no workaround for this I've to live with it as long I use the program.

If PhotoLab restricts you to having to do that, I would use something else, seriously.

[george013]

George, I deleted my previous post which may not apply to your experience.

You said "Why do I get a sharper image on my screen when enlarging above 70% and less sharp when at 70% or lower?"

By "enlarging" did you mean zooming into the review image?

If so, it could be that DxO applies a different re-sampling above a certain zoom level and/or applies sharpening below a certain zoom level. I am no fan of DxO ...

You said "How can I resize the image and only then apply sharpness to it?"

I would open the image and immediately resize it before any other editing. Then when you come to sharpening (normally done last) do it at 100% zoom.



If PhotoLab restricts you to having to do that, I would use something else, seriously.

I mean zooming, the difference between windowsized and 100% and anny other percentage. There is a hard point at wich the picture is getting sharper and that is above 70%.
Resizing can only be done at saving time.
Now i have to figure out if i have to juge the image at +70% or below. And just guess the sharpness of that resized image.
What i see at lower percentages is not the result of my editing.

George

[xpatUSA]

I mean zooming, the difference between window-size and 100% or any other percentage. There is a hard point at which the picture is getting sharper and that is above 70%.

Thank you, now I understand. If you open a file and without any editing zoom in and out, the sharpness of the review image changes at some point while in PhotoLab. To me, that would be unacceptable because the original image has not changed at all.

In (free with donation) FastStone Viewer you can select any view re-sampling algorithm you want and additionally select smoothed or not-smoothed! No jumps that depend on amount of zoom.

Resizing can only be done at saving time.

That is equally unacceptable to me.

Now i have to figure out if i have to judge the image at >=70% or below. And just guess the sharpness of that resized image.

I don't think anyone here would be content to do that!

What i see at lower percentages is not the result of my editing.

I understand.

Based on what you have told us, I would dump PhotoLabs and try something else.

[Manfred M]

George - I suspect Ted is onto something here.

PhotoLab is a raw converter and was never designed to produce output the way you seem to be looking at it. Some things suggest it might be moving in that direction (as Lightroom and Capture One have), but it is not there yet. This is a product that converts raw data into image data for further work in a pixel based editor, so resizing and operations other than input sharpening are not really something one should be doing here. Those are operations that people normally pass on to Photoshop or similar tools.

Where this product excels in is in correcting lens distortion, chromatic abberation and in noise reduction. It does these tasks far better, in my opinion, than Adobe Camera Raw / Lightroom. I turn to it when working with problematic raw files where ACR / LR just don't get the job done to the level I expect.

[xpatUSA]

George - I suspect Ted is onto something here.

PhotoLab is a raw converter and was never designed to produce output the way you seem to be looking at it. Some things suggest it might be moving in that direction (as Lightroom and Capture One have), but it is not there yet. This is a product that converts raw data into image data for further work in a pixel based editor, so resizing and operations other than input sharpening are not really something one should be doing here. Those are operations that people normally pass on to Photoshop or similar tools.

Where this product excels in is in correcting lens distortion, chromatic abberation and in noise reduction. It does these tasks far better, in my opinion, than Adobe Camera Raw / Lightroom. I turn to it when working with problematic raw files where ACR / LR just don't get the job done to the level I expect.

Thanks for the clarification as to what PhotLab actually is - I didn't know. George, please note.

My own camera's converter is excellent at that BUT users still whine about it not doing editor-esque things, e.g. "why can't I ..." crop, straighten, re-size, barrel-roll, etc., ad nauseam.

[george013]

George - I suspect Ted is onto something here.

PhotoLab is a raw converter and was never designed to produce output the way you seem to be looking at it. Some things suggest it might be moving in that direction (as Lightroom and Capture One have), but it is not there yet. This is a product that converts raw data into image data for further work in a pixel based editor, so resizing and operations other than input sharpening are not really something one should be doing here. Those are operations that people normally pass on to Photoshop or similar tools.

Where this product excels in is in correcting lens distortion, chromatic abberation and in noise reduction. It does these tasks far better, in my opinion, than Adobe Camera Raw / Lightroom. I turn to it when working with problematic raw files where ACR / LR just don't get the job done to the level I expect.

I know Photolab is a raw converter. That's why I purchased it. But every photo editing program, including converters, is based on a visual judgement. I just can't explain that 70% boarder and can't ask DxO either. Only when I'm able to contact them in some way.
Resizing is as far as I know also part of other converters. Not sure about that. CaptureNx2 had easy resizing, and the U-point technology. One of the reasons to choice DxO. I don't care about the Nik plugins.
Let's wait.

George

[proseak]

Try here -

https://support.dxo.com/hc/en-us/art...e-PDF-version-

[xpatUSA]

Resizing is as far as I know also part of other converters.

Not mine.

[odds]

...CaptureNx2 had easy resizing, and the U-point technology. One of the reasons to choice DxO.

Functionality dubbed Color Control Point was added to the most recent Capture NX-D. I downloaded it, but I have not yet installed it to see what it actually provides. https://nikonrumors.com/2018/09/13/n...-feature.aspx/
--
Odd S.

[george013]

Functionality dubbed Color Control Point was added to the most recent Capture NX-D. I downloaded it, but I have not yet installed it to see what it actually provides. https://nikonrumors.com/2018/09/13/n...-feature.aspx/
--
Odd S.

I just downloaded and installed it. Feeling different but looks good. Thanks.

For that 70% question https://support.dxo.com/hc/en-us/art...h-DxO-PhotoLab. I don't know yet what that means.

George

[odds]

...For that 70% question https://support.dxo.com/hc/en-us/art...h-DxO-PhotoLab.

That page says:

"OpenCl is used for export as well as when calculating RAW display when the zoom level is above 75%. This is true only if our benchmark has determined that OpenCL performance is superior to CPU processing."

OpenCL (Open Computing Language) can sometimes do calculations faster than the main processor (CPU) by involving more processors available via the OpenCL framework. For example involving a hefty GPU (graphical processing unit) sitting on a graphics board with lots of memory and otherwise idle processor cycles to burn.

Applications may feed a test-calculation via OpenCL to benchmark it to see if it actually performs better than just using the central processing unit in a straight forward manner. Capture One has a setting where the user may set OpenCL usage always off, always on or auto. Capture One will use all GPUs (think multiple graphic cards on one computer) to boost speed.

DxO will apparently use OpenCL for export (calulations) and when calculating RAW display when the zoom level is above 75%. if OpenCL proved to be faster when bechmarked, that is.

You may perhaps like this web page: http://everything.explained.today/OpenCL/

Beware of the "framework battle" as indicated by Apple's move explained here: https://www.imore.com/opengl-and-ope...d-macos-mojave

--
Odd S.

[george013]

That page says:

"OpenCl is used for export as well as when calculating RAW display when the zoom level is above 75%. This is true only if our benchmark has determined that OpenCL performance is superior to CPU processing."

OpenCL (Open Computing Language) can sometimes do calculations faster than the main processor (CPU) by involving more processors available via the OpenCL framework. For example involving a hefty GPU (graphical processing unit) sitting on a graphics board with lots of memory and otherwise idle processor cycles to burn.

Applications may feed a test-calculation via OpenCL to benchmark it to see if it actually performs better than just using the central processing unit in a straight forward manner. Capture One has a setting where the user may set OpenCL usage always off, always on or auto. Capture One will use all GPUs (think multiple graphic cards on one computer) to boost speed.

DxO will apparently use OpenCL for export (calulations) and when calculating RAW display when the zoom level is above 75%. if OpenCL proved to be faster when bechmarked, that is.

You may perhaps like this web page: http://everything.explained.today/OpenCL/

Beware of the "framework battle" as indicated by Apple's move explained here: https://www.imore.com/opengl-and-ope...d-macos-mojave

--
Odd S.

I mean how DxO deals with a raster image. In mine assumption that raster image is created in memory and just send to the screen. What does it calculate? I know for edits as sharpening and maybe noise a high zoom ratio is requested. But only when using that function. Afterwards I can zoom whatever I want. Unless it's working with layers internal and the resulting raster image has to be calculated on the fly.
On this pc DxO is using the cpu, it's faster. On my other pc it's using OpenCl.

I just mailed DxO for the third time. I'm getting an automated answer back immediately. That's all.

George

[odds]

...What does it calculate?

I read you question to be about why/why not OpenCL. Sorry, I do most likely not understand your question. The static raster image model does not fit the way I think of the processes involved.

--
Odd S.