Why are the previews of Capture One Pro of such low quality?

Started by The View, June 27, 2017, 04:14:51 am

DP

first of all - we can't use sharpening... so I switched sharpening off in both ACR and C1 - even that does not make things eq as there are demosaicking involved... and I make both zoom @ the almost ~same ratio - 33% in this case (again that does not ensure pixel by pixel equality in case of different converters)... to my eye (excluding colors and contrast - different converters, etc) - both are the same in terms of preview quality...


bcf

I agree that in your example both previews are of similar quality.

To make things more comparable, I have also disabled sharpening completely. Both images are "as is" (default settings), except for the disabled sharpening. Even in this case I find a marked diféfrence in "crispness" of the preview.

I am at a loss to explain the difference between our findings…
-- Bernard, Lyon, France

DP

Quote from: bcf on December 20, 2018, 09:05:29 am
I am at a loss to explain the difference between our findings...


stupid question - did you rebuild preview to use 640px ? may be it is still interpolating down from some bigger ones ? do you care to share the raw file (this one or something of no value) so that I for example can try on my notebook ?

bcf

Yes, I did rebuild the previews after setting them to 640px in the Preferences. Although it makes little difference: the preview is rebuilt (I can briefly see it being rebuilt), but the result is no better (or only very slightly) than when my previews were set at 1920px.

Do you agree that in the screen copies I posted there is a difference? I would be very interested in your findings.

You can download the raw file here:
<link removed>

Of course all rights are reserved for any usage other than testing the raw processing...  ;D

-- Bernard, Lyon, France

DP

Quote from: bcf on December 20, 2018, 10:13:04 am
You can download the raw file here:


I will try now to compare it on my notebook between ACR & C1

DP

again - I do not see any noticeable acuity-type differences - both @ 33% - zeroed sharpening and NR in both programs and tried to equalize WB and contract

see 2 screenshots


bcf

Now that's interesting! I agree with your findings, there are no noticeable acuity-type differences.

But you do mention "@ 33%" - and this may be the important point. If I view my pic at 33% in C1, I get a good preview, with the same acuity as what you get. But if I go down to "to fit", then the preview is blurry again, even though the difference in size is minimal.

Do you find the same?
-- Bernard, Lyon, France

DP

Quote from: bcf on December 20, 2018, 11:18:05 am
Now that's interesting! I agree with your findings, there are no noticeable acuity-type differences.

But you do mention "@ 33%" - and this may be the important point. If I view my pic at 33% in C1, I get a good preview, with the same acuity as what you get. But if I go down to "to fit", then the preview is blurry again, even though the difference in size is minimal.

Do you find the same?


fit is complicated because you also need to match the size of the visible image in 2 different applications with different screen UI layouts... I might try... I use 33% as the lowest where I can still "equalize" the zooming(out) between ACR and C1 on my LCD.

bcf

Don't try to compare C1 and LR for the "fit" setting. What I mean is that using C1, I see a visible loss of acuity when I switch from 33% down to "to fit". I guess the "to fit" preview is less optimized than a fixed magnification such as 33%

The problem for me is that I like the "to fit" setting, with 90px margins, because it allows me to have a better "overall" view of the image. If I go to 33%, then I have to scroll the image vertically some to see the top or bottom.
-- Bernard, Lyon, France

DP

Here is what I did

1) made ACR window small enough so that "Fit" in ACR will be 25% exactly (at least as reported by ACR)

2) made C1 window small enough so that "Fit" there will be as close to 25% (but most probably not exactly 25%) - hence it is ~25% in C1 vs 25% in ACR

see both screenshots


DP

Quote from: bcf on December 20, 2018, 12:52:48 pm
Don't try to compare C1 and LR for the "fit" setting. What I mean is that using C1, I see a visible loss of acuity when I switch from 33% down to "to fit". I guess the "to fit" preview is less optimized than a fixed magnification such as 33%


see my post above - there ACR 25% is vs "Fit" (not exactly 25%, just close to it) in C1

bcf

Am I dreaming or is the preview in C1 now _better_ than the one in ACR?
-- Bernard, Lyon, France

bcf

This is what I mean : 2 screenshots from C1, one with the viewer set to "to fit", the other at 33%.
I find the view at 33% sharper than at "to fit", even though the size difference is minimal.

My screen is a 27" (2560*1440).

I must say I don't know what to conclude...
-- Bernard, Lyon, France

DP

Quote from: bcf on December 20, 2018, 01:04:42 pm
Am I dreaming or is the preview in C1 now _better_ than the one in ACR?


I am not using "Fit" size in either program to evaluate the acuity-related details - only WB/brightness/contract/clipping of various kinds (overexposed/underexposed/gamut) in general over the frame ... so honestly for me (my LCD is PA242W still) even if C1 is worse it is irrelevant... details like demosaick/sharpening/noise/chromatic aberrations/clarity/etc I check @ 100% or use loupe

DP

I tried to downsize C1 window while keeping "Fit" to make sure that viewing at odd sizes less than 25% - like 20-something or high teens is not impaired... it is not... at least on my computer + lcd + whatever settings I have...


DP

when I drag with mouse pointer C1 window resizing it I can see momentarily the loss of acuity, but then when I stop and spinning ball/star showing that C1 does redraw (from cached raw I assume) finished & disappears the acuity is back to expected...


DP

with C1 a wild guess might be

1) disable GPU - see what happens - enable GPU - see what happens

2) play with GPU drivers (upgrade/downgrade)

3) manually delete preview files from the disk - I am using sessions always, no issues to do this there - not sure how it is handled with catalogs (while keeping previews settings @ 640px) and see what happens next when they are genuinely rebuilt new

4) disable "enable recipe proofing" if you have it

bcf

Thank you for this information. I'll try to digest all this and explore some more...
-- Bernard, Lyon, France

David Grover / Phase One

Quote from: bcf on December 20, 2018, 02:10:35 pm
Thank you for this information. I'll try to digest all this and explore some more...


Also work through.

https://www.phaseone.com/en/Search/Article.aspx?articleid=1721&languageid=1

And if you didn't supply screenshots to support.  Please do so.  :)
David Grover
Business Support and Development Manager, Software.

bcf

Thanks David.
I tried the various steps mentioned in the document :
- Disable the use of OpenCL by C1 (in the Preferences), restart C1: no change.
- Reenable the use of OpenCL, quit C1, delete relevant files to force revuilding of OpenCL kernels (I'm on a Mac), restart C1: no change.
- OpenCL is indeed used, as shown by the manoeuver described in the document (using the focus mask).

I would say OpenCL is not the culprit.
-- Bernard, Lyon, France