Started by The View, June 27, 2017, 04:14:51 am
Quote from: Dinarius on July 24, 2017, 11:35:47 amI've always had the same experience, but learned to live with it.What I see in Photoshop is a tack sharp TIFF, with smooth tonal variation, and less contrast and saturation.What I see in C1 is something softer, with more contrast and a little more saturation.C1 is set to its Output Recipe (always Adobe RGB/16 bit TIFF) and Proofing is turned on.There's been a lot of talk in this thread about what one sees at 100%. Yes, what I see at 100% is identical in both C1 and CS6.But, when I'm editing, I want to view the overall image most of the time, to get a "feel" for what's happening as I edit. I don't want to be making global adjustments at 100%.And, for whatever reason, I'm not seeing that in C1 when viewing the full image (as opposed to viewing at 100%)This problem is quite evident when viewing my 5D Mark IV files, and extremely so when viewing my Sony RX100 files. The Sony files are transformed after saving and opening in CS6.In C1, Edit/Preferences/Image is set to the default 2560. Colour is Perceptual. Open CL is set to Auto and being used by C1.I would love someone to tell me that I have screwed up a setting, and my issue can be solved. Thanks.D.
Quote from: BartvanderWolf on July 27, 2017, 09:24:28 amI have difficulty understanding what some are saying that they observe. An example of two screenshots, one from C1 and one from the TIFF conversion in PS, might be helpful for a meaningful discussion.I've tried viewing an existing TIFF in both C1 and PS at a fit screen zoom-level, and they looked almost identical (except for minor resampling quality differences). In C1 I had to turn off all my default style changes, in Contrast, Saturation, Clarity, Sharpening, etc. The general lack of differences that I observed, suggests that it is probably not the resampling that produces the effects (e.g. blur, contrast) mentioned by others. If true, then that would leave the generation of the preview image when viewed at a reduced (fit screen) size. If that's the case, then perhaps the generation of a larger (than fit screen) size previews might help? It would slow down imports and use more space to store previews. It's all a bit hard to say something meaningful about it without actual comparison examples of (C1 preview versus PS display of the resulting TiFF, both at roughly 'fit screen' resolutions) screen shots which demonstrate the observations. It could also be something else, like Retina display settings, who knows?Cheers,Bart
Quote from: Dinarius on July 28, 2017, 07:31:18 amWith a file open to "fit in screen" in C1, save it as a TIFF and open it in Photoshop (in my case CS6).Now toggle between the two.Are you saying that you don't see a difference?I'm Windows 10 Pro x64, and unless there's some setting in either program (C1 / CS6) that I should be changing, I do see a difference - as I outlined above.
Quote from: The View on June 27, 2017, 04:21:47 pmThe image size is set to 2560 pixels - which is the resolution of my 27" monitor.But compared do DPP, sharpness and detail is much reduced. I have experimented to set it to an even higher pixel size, but this did not bring any advantage.
QuoteCould it be that DPP has a very high contrast setting out of the box? The preview images is just so much crisper while the C1 preview is "soggy" and "softish", inviting the use of contrast and detail sliders.
QuotePS: hardware acceleration (us open CL) is set to "never" for both display and processing. It was a recommended setting a while ago. Is this still the best setting? I'm on a 2012 retina MacBook Pro with 16GB ram.
Quote from: BartvanderWolf on July 28, 2017, 09:21:31 amHi,What I see is a very slight difference, due to different resampling methods (and slightly different zoom ratios due to workspace layout) and a different CMS. What I do not see is blurry images, or very different contrast, or much changed colors, that would qualify as low quality. As said, they are not 100% identical (C1 is slightly darker, or PS slightly lighter) but very close, despite obviously different technologies being used. If I would not have labeled them, you would most likely not be able to tell which one is from what application.That's why I asked for an example from those who do see issues, in order to ferret out whether it might be caused by certain settings (e.g. wrong settings in the export recipe or the display settings).Anyway, I've spent some time and produced the attached side-by-side screenshot comparison, which looks quite similar (not optimal due to conversion from display profile to sRGB), and without output sharpening or the other refinements in Photoshop that I normally would do. But for sharpening itself, one must use 100% zoom (or larger) anyway, so this is just for getting an overview of the image.So, slightly different yes, low quality no.Cheers,Bart
Quote from: BartvanderWolf on July 29, 2017, 06:52:18 amDPP uses better/additional processing for downsampled / fit to screen previews. So it's not that C1 is particularly fuzzy, it's that DPP is particularly sharp. I assume that C1 just downsamples (like LR does), without compensation for the resampling. Resampling by definition softens the image (depending on the filter used), otherwise the images would be riddled by aliasing artifacts. Photoshop's downsampling is somewhat crude (quick and dirty). Adding additional sharpening for display, would add processing time. In theory, C1 could add that as an option, just for downsampled display, or for previews. I've experimented with such post-resampling sharpening in the Resampling script based on ImageMagick that I developed in cooperation with Nicolas Robidoux here. My addition resulted in an adaptive deconvolution filter (a Difference of Gaussians filter) that would be used for a deconvolution which could easily be optimized to run faster if encoded in a lower level programming language.As mentioned before, especially sharpening should only be done at 100% zoom (or larger for detail inspection). Capture One Professional nowadays has excellent sharpening Proofing for that purpose. Contrast is hard to nail at reduced sizes, because that involves eye adaptation that subtends a small angle of 1 degree, and a phenomenon which is also known as simultaneous contrast. A slightly sharper fit to screen preview only solves a very small portion of that issue.That is only necessary if hardware acceleration causes errors. I have it on auto, without issues.Cheers,Bart
Quote from: The View on August 22, 2017, 05:21:53 pmThe comparison I am using is what I get when I output an image. And the output I get from C1 - either to Tiff and to open in Photoshop, or directly to jpeg for first stage proofs - the output is very different from what I see in a preview. Contrast and general appearance are so obviously different that I am astonished that this hasn't been noticed more.
QuoteDPP is very sharp in the preview, but so is the output.
QuoteThe foundation of all digital image editing is that what you see is what you get. And the output of C1 is several degrees better than the poor preview. The difference is so striking that I simply can't rely on the preview and I use only minimal adjustment in C1 because of this. Sure, I could zoom in 100% - but edits need to be done seeing the whole image to really see what one is doing. Peeking at a crop won't do it.There is an absolute need to work on this - all those extra features and editing features cannot be used with a specific outcome in mind because the full view is so off.You remember, Photoshop had this problem, too in earlier versions like CS3. You could get a really blurry picture with percentages that were off the 25% 50% 100% sequence, and 50% used to be better than 25%. Now you can get a good representation of the image at almost any percentage. Photoshop's display/preview coding has gotten extremely good.Wouldn't be achieving such a reliable preview a great goal for C1 11?
Quote from: BartvanderWolf on August 23, 2017, 11:05:36 amMaybe because others do not see such 'very different' results? Hard to say without an example from what you see. The differences that I experience are rather small.Indeed, but they use additional processing. I remember that the earlier versions had an option that the user could turn on to get a higher quality display. I assume that the current version has that switched on by default, and it no longer offers such an option. They probably assumed that faster hardware made it less of an issue to do the extra processing with a reasonable time delay.I wouldn't object if they offered an option for sharpened zoomed-out display. A very simple deconvolution on down-sampled images is not too complicated IMHO, but it is an additional step that would slow down image display as images and display resolutions grow larger and larger as technology progresses.Cheers,Bart
Quote from: David Grover / Phase One on July 28, 2017, 06:29:07 amThere should be no reason why you couldn't adjust comfortably at fit screen. ... Or use the excellent proofing feature.
Quote from: DanFreck on August 26, 2017, 07:21:30 pmA picture is worth a thousand words, so
Quote from: DanFreck on August 26, 2017, 07:21:30 pmCan anyone help with this?
Quote from: TeeKay on August 26, 2017, 10:44:19 amThere are a number of reasons why the standard "fit screen" rendering of C1 is suboptimal.It is too soft. This makes it hard to ascertain whether a shot is usable or not without zooming in and out all the time. Not everyone likes to use the focus mask feature. The general softness of the image also makes it impossible to enjoy ones images while browsing through them. It does not make sense to be forced to either zoom in at least a couple of steps or use the proof feature in order to get a crisp rendering.Noise can look vastly different.Probably, due to the change in noise rendering characteristics, colour tints can be introduced in the previews that are neither visible when zooming in, nor in the exported images. Its been a while since I worked with images of this kind, so perhaps the latter issue has been addressed in the meantime.I realise that downscaling comes with its challenges and that of course noise looks different when downsampling a large image. I am familiar with the notion of noise levels being dependent on spatial resolution. I understand why a normalized 8MP image can have a higher dynamic range then a 24MP original (-> DxOMark). However, it is also clear that the fuzziness of C1 previews are not owed to insurmountable downscaling problems. The proof that useful renderings at the "fit size" are possible arethe exported images at such (low) resolutions.the rendering when the "proof" option is turned on.the rendering when one sets the preview size in the settings such that one forces C1 to always generate the preview afresh.It seems that C1 -- if a preview image of sufficient size is available (has been cached) -- uses that preview source to downsize it to the "fit size". This process yields a different result compared to when C1 renders the image at the "fit size" directly from the RAW data. In my view, that is the problem with C1's previews which clearly could (and should) be more useable, obviating the need to have the proofing option turned on with the distracting "Proofing" label in one's sight. I believe the preview rendering got better in V10 (compared to V9) but it is still not as good as it should be, AFAIC. Activating the "proof" option is a workaround, but I don't like to see the distracting "proof" label all the time and it also should not be necessary to regenerate from the RAW data all the time. All it would take to show crisp previews very quickly is to store more cached versions (at various sizes) or a better downsizing from a larger preview. For the downsizing, the user could be asked which sharpness parameters to apply (or parameters adapted from those in the proofing options could be used).It clearly it is not a solution to zoom in at 100%, or even just a couple of steps (the latter often being enough to force a proper generation which results in a crisper rendering). I understand that the preview has to be a trade-off between accuracy and something that can be produced/shown quickly, and I can accept that as an explanation. Explanations alluding to downsampling challenges in principle are, in my view, not acceptable for the aforementioned reasons.
Quote from: DanFreck on August 28, 2017, 06:52:40 pmThank you for the vindication guys. This really is a "core and crippling problem".I've been discussing this over at the phase one user forums. I ended up refining my examples down to the bone, to remove any comparison issues, and make everything equal. Here's what I put there:I made a preview without any adjustments. This way any edits I made play no role whatsoever. Took a screenshot, again from within Capture One:http://scenicdesktops.com/test/CapturePreview.pngThen, I took that screenshot, of exactly what's on my screen, and cropped out just the preview part. It ends up being 1927 x 1286:http://scenicdesktops.com/test/CapturePreviewCropped.pngThen, in capture one, I exported the image at this same resolution, 1927 x 1286. That way the output matches the exact dimensions of what I see, on screen, in capture one.http://scenicdesktops.com/test/CaptureOutput.pngNow compare those last 2 links. No difference in resolution, no adjustments, just pure comparable quality differences between two 1927 x 1286 pictures, generated by capture one in different situations. There's an obvious difference between what you see in capture one, and the image itself being worked on, which could undeniable look better (as the output demonstrates quite clearly). Look at the trees, mountains, and logs at the bottom right. The same program produces a mushy, cruddy looking picture at 1927 x 1286 in-program, while at the same time producing a better, higher quality picture at the same resolution of 1927 x 1286 when exported.I will try and create a support case with Phase One. I'm moving away from Aperture, since it's been unsupported for a while now, and my desktop PC is faster and easier to edit with anyway. I expected this program to be better, and gave it the benefit of the doubt at first. Now I'm turned off, because this non-destructive raw program can't seem to even make decent in-program images to work with. This is face-on obvious stuff, which I noticed right away the first time I used capture one. How are people not up in arms about this? How old is this program?
Quote from: myotis on September 07, 2017, 07:17:01 amTom Fitzgerald has just published a suggested workaround for thishttp://blog.thomasfitzgeraldphotography.com/blog/2017/9/capture-ones-preview-problem-and-how-to-get-around-it
Page created in 0.054 seconds with 15 queries.