Some initial observations. PL8+FP7+VP4/Win11, i7 14th gen, RTX 4070, 32GB RAM, 2 M.2 disks, Nikon cameras here. Switched to PhotoLab some 10 months ago from old Lightroom. Just finished in PL8 a 500 photo project (concert in badly lit room, some very noisy images) and 900 snapshots for private use (landscapes, nature, streets, people, mostly in good light).
PL7.9 → PL8.0 upgrade went smoothly. My presets, preferences, main and modules databases were carried over automatically (I have entered the license key during installation, if it matters). The preview and thumbnail caches (10+ GB in my case) were NOT migrated, probably because it would require re-rendering to be in sync with the new engine, which could take days. EULA and 37 days offline limit are still the same.
Loupe:
At least 50% of my images are taken at very high ISO, so inability to display live DeepPRIME results was my first concern, when I switched to PhotoLab (PL7). It took me several months to “learn” DeepPRIME in PL7, with my preferences changed several times after dealing with new types of photos, noise level, particular subject properties, color temperature, and even the lens. I hope I’ll learn DPXD2S much more quickly because of previous experience AND the loupe. That said, I would like to have two or more loupe dimensions to choose from (small ‘±’ button could be used in loupe for more than two dimensions). I have quite strong hardware and I can choose to wait, say up to 7 seconds, for the final preview to appear. But even the current implementation made my workflow noticeably smoother and faster.
Still missing: Full preview on demand. Generally I look for emotions or harmony more than for details. Hence for most of my work I use the whole image displayed on a 4k monitor for overall perception judgment. I look at 100% or 200% only if the image is exceptionally good and worth pixel-peeping, or if the image resized to 4k monitor looks suspicious in terms of focus or artifacts. Given the fact that some other corrections are displayed only at 75%+, and I started to use 45mpx input (which uncropped is well below 75% treshold on 4k), the full correction preview on demand is still missing. Loupe shows only 100% resolution or more, so it’s not meant for this purpose, even if it could be enlarged. Currently I use the workaround given by @Wolfgang recently somewhere in this forum (export + ‘Compare’), but full preview on demand would distract less.
DeepPRIME XD2S:
- Some images of people looked over–sharpened due to “fake” microcontrast added in PL7 DPXD, especially for things like hair, which was seen only after export. It seems less of a problem in PL8 and I can use loupe to judge. Some may complain that XD was “sharper”, but imho XD2S is more natural. These observations are valid only for very noisy images.
- Maze and some moire patterns seem to be handled much better, but it’s my preliminary opinion based on only several cases. Some moire are impossible to fix anyway by any software now, and perhaps in the future. Some of them seem to be impossible to detect and fix by humans using pixel editors.
- There’s less of what I call ‘micro-maze’ compared to XD (noticed only when pixel-peeping). Maybe it has something to do with deconvolution kernels, probably used by some lens corrections and perhaps joint demosaicking and denoising – just useless speculation.
Hard to say what’s missing here, as we are at the edge of technology. Surely, there are known algorithms using object recognition to better deal with noise, but they are too energy greedy for practical home use and are even more prone to artifacts. Maybe they are/will be offered in a cloud, like some AI Microsoft services in the future, after restarting the Three Mile plant they bought recently. But that’s getting crazy…
Tone Curve:
You can use ‘level indicators’ to precisely set coordinates of anchor points. This is of particular importance for deep shadows, where PL7 ToneCurve usefulness is questionable, IMHO. Few months ago I asked for ‘DeepBlacks’ selective tone slider in PL7, but now one can emulate “DeepBlacks selective tones slider” to some extent using the ToneCurve and level indicators. Also the histogram at the curve background gives some hints for choosing anchor points. I almost never used ToneCurve in PL7 but this may change in PL8.
Filmstrip performance:
I often work on projects which start with 3000 images, so I appreciate the difference. ‘Select all’ is much faster. Usually such improvements come at a price, but so far no bad side-effects seen (using it few days only, though).
Compare tool:
Nice to have new functionality, although I don’t use it too often. Most of the time I use “press and hold” the Compare button.
Missing: switch to synchronize magnification and navigation, while comparing side by side two images coming from the same source.
Hue mask:
I used it for fun on one photo to change the shirt color. It worked really well but the subject was too easy to make a general, reliable opinion. I rarely use local adjustments anyway (except for retouch, but that’s generally considered not to be LA) and only if they don’t require too much time. Most of it is adjusting slightly face exposure, which is usually easy.
Missing: Some combination of luminance and hue mask would probably be useful for me. Maybe some polygon tool to limit the mask area quickly (like in Nik?).
Missing: Better face recognition in SmartLighting. It seems there’s no improvement here. But that’s a rather minor issue for me, as I often add some other areas anyway.
I’ve lost faith in DxO’s quality control after recent huge memory leak during exports on Mac. So far I haven’t experienced any major issues with PL8.0, typical for new versions. It looks more like a minor release. I’ve documented elsewhere in this forum two minor bugs with PL8.0, one with tone curve and one with compare side by side. These are obvious and should have been detected by DxO acceptance tests, so my doubts about DxO QC still remain.