thedigitaldog  wrote Knicker   wrote What was not clear here (until I dropped into the discussion today) was whether interpolation algorithms apply to zoomed-in previews. Until I mentioned it, nobody questioned the "Answer" that zoomed-in previews (in LR's library module) should  be interpolated. It wasn't clear to you. And you're of course, still speculating about interpolation algorithms with zoomed in previews in any LR module. Some here can ask specifically what's going on, to an Adobe engineer via email. Not really necessary..... The running editorializing isn't called for and isn't appreciated. This will be my last post here until you or one of the other Adobe insiders report back on what Adobe tech/engineering is advising. Here's what I predict they will confirm: In PS, some form of (probably bicubic) interpolation is applied when zooming out at odd percentages. In PS, nearest neighbor or something similar is used when zooming in to preserve the pixel structure appearance. In ACR, some different form of interpolation is applied when zooming out at odd percentages. This is one reason why sometimes users are confused by the difference in appearance of zoomed out images in ACR compared to the same images zoomed out in PS. ... View more

thedigitaldog  wrote Knicker   wrote To recap: it's not changing the interpolation preference setting in PS that's causing a change in the interpolation method when zooming out on my iMac. It's changing the GPU on/off status that's causing the change. I've now confirmed (on my iMac, at least) that the zoom-out interpolation algorithm is fixed to whatever it is in PS regardless of the preference setting. No. As explained, the differences is due to bugs with GPUs. I don't know how you keep missing that fact but perhaps this will help: On my Mac, GPU on or OFF produces and IDENTICAL preview at 200%. BECAUSE as I've attempted to illustrate to you over and over again, my system doesn't exhibit GPU bugs: You're still mixing two separate issues. When zooming IN, my iMac behaves correctly like yours does - ie., no difference in the display whether or not the GPU is on and no interpolation applied. What I've confirmed this evening after reviewing my tests from last night is that, on my iMac, when zooming OUT in PS there is a visible difference in the interpolation applied (see my crops in my prior post above from the DPR studio scene). Your 200% example above proves nothing about whether or not your Mac also exhibits the same "bug" as mine does when zoomed out. I suggest that you give it a try and report back. Bear in mind, that it's the GPU-off variation that looks rather wrong on my iMac, not the GPU on variation. That's what's odd about it. ... View more

Well, for whatever reason, the bilinear interpolation on zoom setting is how my version of FRW was defaulted. Until you mentioned above that it's configurable, I wasn't even aware of it and hadn't tried to look for it, let alone change it in the preferences. (I confess that I never made it to page 113 in the user manual.) Regardless, this is off-topic to an Adobe forum. PM me on DPR if you and Iliah want more info from me about my FRW configuration. And, by the way, interpolation algorithms ARE applied to zoomed-out previews in PS/LR. That's not debatable. What was not clear here (until I dropped into the discussion today) was whether interpolation algorithms apply to zoomed-in previews. Until I mentioned it, nobody questioned the "Answer" that zoomed-in previews (in LR's library module) should  be interpolated. ... View more

thedigitaldog  wrote Knicker   wrote I remember the first time I tried the zoom function in FastRawViewer, I was really thrown for a loop. Zooms past 100% are interpolated/smoothed. I still scratch my head about why they programmed it that way. A question for Iliah, I suppose, in another time and place... I don't think so. There IS a major difference in rendering (contrast and saturation). To be expected. But you should ask him about this concept of interpolated and smoothed, I think he'll tell you otherwise because it makes no sense to me to do so. The differences are seen below on a system I know previews correctly pretty close if you discount contrast, saturation etc: 200% in each. Go to 1000% in FRW (the max). Do you see any pixelation at all? I don't on my iMac, but at 1000% you should definitely see the pixel structure. ... View more

thedigitaldog  wrote Knicker   wrote Try setting the preference to Bicubic Sharper in preferences, close/reopen PS, zoom out to an uneven percentage and do a screen grab. Rinse and repeat with Bicubic. The screen grabs should show a noticeable difference. It absolutely doesn't on this end. No matter the interpolation algorithm in preferences, the zoom uneven percentage or not, the image previews identical no matter how set, no matter how many times I relaunch Photoshop. It makes absolutely no sense that Adobe would make the product behave this way anyway and they didn't. Now that isn't to say a bug somewhere may produce this utterly undesirable effect. Fortunately for me, it doesn't. Again, it makes absolutely no difference in what you see zoomed on-screen based upon the resample algorithms in Photoshop. And Adobe makes it perfectly clear what those algorithm's affect: new pixels based on the color values of existing pixels. You're right. I was trying a number things last night and mislabeled my findings and accidentally attributed the change I was seeing to the default interpolation setting in the PS preferences, when in fact it was due to toggling the GPU on/off in the PS preferences. To recap: it's not changing the interpolation preference setting in PS that's causing a change in the interpolation method when zooming out on my iMac. It's changing the GPU on/off status that's causing the change. I've now confirmed (on my iMac, at least) that the zoom-out interpolation algorithm is fixed to whatever it is in PS regardless of the preference setting. It's also fixed in ACR. What remains true, however, is that the zoom-out algorithm used in PS is different from the one used in ACR. And, of course, this is a different set of issues than the zoom-in behavior settled earlier in this thread. What's odd and kind of depressing is that my generic iMac using only it's embedded GPU is displaying visibly different zoom-out behavior in PS when the GPU is on compared to when it is off. The gpu off setting looks worse. Weird. If anything, you'd expect the GPU turned on to be the source of any problems, not the reverse. Here's an example of what I'm talking about (note the strange repeating box like structure introduced by the gpu-off version of the DPR test scene (opened from ACR into PS) and displayed at "Fit on Screen" size (approximately 23% on my 27" iMac): The crop on the top is GPU-on and the crop on the bottom is GPU-off By the way, if you're interested in a good test image to use for quickly comparing different interpolations for downsizing or zooming out, try the one below. The grid will look dramatically different with different interpolation algorithms applied when downsized/zoomed out to odd percentages: ... View more

thedigitaldog  wrote Knicker   wrote The referenced 2017 thread discussed how, when zooming out, there may be differences in the appearance of the displayed image between the Library and Develop modules. That's correct. It's also true in Photoshop vs ACR, depending on how Photoshop's image interpolation preference is set in the Preferences > General tab. I don't believe that's at all correct either; this setting in Photoshop's preferences has no role over previewing data on-screen when zooming, nor do I see any difference doing so. It's used for actually resizing (interpolating) the document data. It's my understanding that the interpolation method used in zooming out in ACR is fixed. You can, however, change the interpolation method in PS itself by changing the setting in preferences. (Like Todd advised for testing LR, you need to be sure to close and restart PS to see the difference). Try setting the preference to Bicubic Sharper in preferences, close/reopen PS, zoom out to an uneven percentage and do a screen grab. Rinse and repeat with Bicubic. The screen grabs should show a noticeable difference. ... View more

thedigitaldog  wrote Knicker   wrote The question is: which way is the unintended/buggy behavior? That should be fairly obvious; the behavior that matches Photoshop. Yes, of course, and it's been "fairly obvious" to me from the beginning of the DPR discussion as well. One of the reasons I don't use LR is because it doesn't always match the behavior of Photoshop. In its efforts to simplify things and protect LR users from some of the complexities of PS, Adobe has created some FUD for those of us used to how PS does things. This didn't strike me as one of those instances, though. I remember the first time I tried the zoom function in FastRawViewer, I was really thrown for a loop. Zooms past 100% are interpolated/smoothed. I still scratch my head about why they programmed it that way. A question for Iliah, I suppose, in another time and place... ... View more

Thanks, Todd, for re-opening the thread and following-up with your own testing! It's interesting that your system's behavior seems to match the OP's - i.e., with the GPU on you get antialiasing applied in the Library module. As an, fyi, on my iMac, turning the GPU on (with full close/restart of LR as you've suggested) does NOT result in smoothed/antialiased zoomed displays. Same is true with the GPU off on my iMac. The question is: which way is the unintended/buggy behavior? My expectation is for all zooms to behave identically and that zooming in should, under all normal circumstances, retain the same pixel structure and just magnify that structure rather than resample/interpolate it. If I read you correctly, you're agreeing with that. The bad news is that would mean that your system (like the OP's) is buggy.   (Better yours than mine! ) ... View more

If the "Answer" was incorrect, misleading or easily misinterpreted in some way, isn't the best place to discuss that in the same thread as the answer post was given? I will defer to the moderator(s) on whether this thread or a new one is the proper procedure. Either way, I sincerely hope the individuals to whom my post is directed see it and are given the opportunity to respond. The goal is clarity on this issue. As it stands now, I don't see how the OP's second question is correctly addressed by the post marked as the "Answer". ... View more

A parallel thread to this one was started by the OP here on Digital Photography Review (DPR). The "answer" post here along with the post by the Adobe staff member (Akash Sharma) were cited in the DPR thread as authoritative. I expect that the "answer" post and this thread in general might also be cited in the future in other threads in this forum (just like the related 2017 thread was referenced above). Therefore, it's important to get things right and to be clear about what's really going on in the Adobe products when the user zooms in and zooms out. As things stand now, there are two statements that are incorrect or, at least, unclear and easily misinterpreted. First there is this statement by Todd Shaner in the "Answer" post: If the Library module looks the same (i.e. pixelated) as the Develop module at >1:1 Zoom view the most likely cause is an incompatible graphics driver. Second, this statement in Akash Sharma's post: Develop module shows the rendered preview of your image but it is best viewed in 1:1 ratio. This means that your GPU processor have some issues and is showing the rendered preview pixelated. As best I can tell both of these claims are based on an incorrect assumption that Adobe uses the same interpolation and resampling process for zooming in on the image display as it does when zooming out. The referenced 2017 thread discussed how, when zooming out, there may be differences in the appearance of the displayed image between the Library and Develop modules. That's correct. It's also true in Photoshop vs ACR, depending on how Photoshop's image interpolation preference is set in the Preferences > General tab. However, something different happens when you zoom in. As far as I can tell, all of the Adobe display modes being discussed here disregard interpolation when zooming in because the whole point of zooming in is to magnify the actual pixel structure at the current pixel dimension settings. Thus, pixelation is exactly what one should expect when zooming in. This is easily confirmed by just going to any image and zooming to 8:1 in LR or 800% in PS. You will see the actual pixel structure of the image magnified and more visible. You will NOT see the pixels interpolated so that edges appear smooth. In this regard, zooming in is fundamentally different from an actual image upsizing/resampling to 800%. When upsizing/resampling through an export in LR or image size change in PS, Adobe is indeed applying an interpolation algorithm. Depending on which algorithm is applied (in PS), you'll get varying degrees of pixelation/smoothing. In particular, it's noteworthy that use of the "nearest neighbor" option will result in a highly pixelated look that's virtually identical when displayed at 100% to what a 800% zoomed display of the original image (pre-upsizing) looks like. With that in mind, the two quoted statements above need to be reconsidered. Images viewed at a zoomed setting in the Library module should normally look pixelated, just as they do when zoomed in in the Develop module. At least, that's what I see (and expect to see) in LR Classic CC when I look (regardless of whether the GPU acceleration is turned on or off). Likewise, "the rendered preview" looking "pixelated" in the OP's zoomed in screen grabs is expected and, therefore, is not a symptom of the GPU having problems. Of the two screen grabs posted by the OP, the pixelated one is the expected on in both modules. The real mystery is why his Library module (not the Develop module) image viewed zoomed in showed antialiasing-type smoothing when the OP's GPU was turned on. It appears that the GPU is stepping on the Library modules zoom function and imposing an antialiasing algorithm to it, but is (correctly) not interfering with the Develop module's zoom function. Whatever the source of the problem, the bottom line is that visible pixelation when zooming in is the expected and correct behavior by which "incompatible graphics driver" issues must be assessed. ... View more