All i can say:
From this photo - DPReview - Studio Scene - Fuji X-M5 - 26MP X-Trans sensor - RAW - ISO : 160. Link: https://www.dpreview.com/reviews/image-comparison/download-image?s3Key=130069f8957940449f3be74d029aa06d.raf
No adjustment, Lens module downloaded, just NR set
Export to JPG - ‘Standard output’ - (default setting, Quality 90).
Pixel by Color count (with this - Pixel Color Counter)
| NR Type | Pixelcount by Colour |
|---|---|
| Standard | 651 611 |
| DP3 | 608 904 |
| DP XD2s | 602 976 |
| DP XD3 X-Trans | 588 480 |
Spoke to an old friend. He has an X-T4 and is a very long time Fuji shooter. I sent him a couple of RAWs and the resulting jpegs. He looked at them and then ran them through DeepPRIME XD PL7 and then downloaded a trial copy of PL9 and used DeepPRIME XD3 X-Trans.
Same result and same conclusion. He then told me that he had looked at PL8 and there wasn’t enough in it for him. As for DeepPRIME XD3 X-Trans. He believes that it removes more detail and does not make the image any cleaner than DeepPRIME XD. He then tried a couple of his own RAWs (higher ISOs), and found similar results.
In his own inimitable way he suggested that the “X-Trans” part is designed to tie Fuji users in whilst other products gain on them. He an excentric sort of guy but he is that sort of guy that when he speaks, people listen.
So far, this isn’t convincing me that there’s a problem. A change with equal settings or default settings, yes. But you haven’t said if you’ve tried adjusting the settings for XD3 and Lens Sharpness Optimization to taste and then compared with PL7’s best settings to see if there’s an improvement. DxO hasn’t committed to preserving the behavior of these tools across major releases; on the contrary, they have documented from the start that the models in use and their biases have changed.
Well you see that a bit of an issue, because what DxO state as “Lens Sharpness Optimization” isn’t really that. It’s “Lens and Camera Sharpness Optimization”.
Unfortunately they insist on providing a solution to the combination. The problem with that is if you use the Fuji X-H2S and a Canon EF100-400 Mk2. Although each are covered individually, as a combination, the lens sharpening is not available. Another reason to look over the garden fence.
I digress. The reason for sending the RAWs to my friend was for him to run some tests. I didn’t even tell him what i thought the problem was. He tested the RAWs with comparable edits, various amounts of DP, with and without lens sharpening. His day-time job used to be a forensic scientist - he’s that sort of guy.
“So far, this isn’t convincing me that there’s a problem. A change with equal settings or default settings, yes.” - So why would equal settings produce a worse image? The idea of DeepPRIME XD3 X-Trans was to improve things for Fuji users, not make the image worse.
My previous post was intended to explain this from DxO’s own release notes and other product literature. They changed the biases toward noise reduction, detail retention, sharpening/microcontrast enhancement, etc. This would explain the behavior you’re reporting. I would expect you can tune this to your liking with the noise reduction settings for XD3. If that’s not the case, then you might get either an explanation or a fix from DxO support by writing to them at support.dxo.com. DxO has their preferences for what is a good balance between noise reduction and detail enhancement. They don’t necessarily match ours.
“They changed the biases toward noise reduction, detail retention, sharpening/microcontrast enhancement, etc.” So the DeepPRIME XD3 X-Trans should be an improvement in noise reduction and sharpening - but it isn’t.
“DxO has their preferences for what is a good balance between noise reduction and detail enhancement. They don’t necessarily match ours.” - from a commercial point of view, providing a product that suit the manufacturer but not the customer is probably going to end in tears. Most companies listen to the customer and deliver, not the other way around. A good balance isn’t the same noise reduction and more loss of detail. Especially when this is meant to be especially for Fuji users.
This problem was apparently reported when PL8 came out and nothing happened. Somewhere on the To Do List, I guess*.*
I have raw files from a Nikon D70, a Nikon D300, and q Nikon D750. I don’t own these cameras anymore, but I still have tons of raw images shot with them. I also have raw images from three Fujifilm cameras: an X100s., an X-T30 (which I sold) , and an X-T5. (my main camera today). I use PR 5 for raw conversion and I have to say that it does a better job than any of the other applications I’ve tried.
I process my RAFs with PR5 (DeepPRIME XD3 X-Trams) and edit the resulting TIFFs (or, more rarely, DNGs) win Photoshop, ON1 Raw, or Luminar Neo, depending on what I am after. What I am no longer using is PL; the last version I have is PL6. While I still want DxO’s raw conversion (at least for the moment),. I am absolutely NOT willing to put up with the way DxO’s releases PL versions before they are ready. Adobe, ON1, and Luminar certainly release versions with far less drama than DxO. Frankly. I don’t understand why the people here put up with this nonsense.
The problem is a financial one. The new PL is released every year at about the same time. So it “has” to be ready. The investors expect to see the money rolling in at this time. So the latest version has to be ready on the said day. They pay “affiliates” to lie through their teeth to convince you to buy it - they are not the only company to do this, but if someone is an affiliate, believe what they say with a pinch of salt.
As for lens adjustment I have found out that my Canon EF100-400 Mk2 needs virtually no adjustment at all. Which is fortunate for me because DxO cannot provide that adjustment as I use it with a Fuji X-H2S. My suspicion is that lens adjustment is overhyped for the better lenses.
As for DeepPRIME XD3 X-Trans - I think this is a marketing ploy to satisfy Fuji users, the results are not as good as DeepPrime XD in PL7. Noticeably.
In short, PL9 offers masking that is some way behind the competition but is likely to improve in PL10 or perhaps before but more likely PL10. Other than that, I see no advantages over PL7 and some disadvantages.
I do like the workflow of PL, the dop file being with the original RAW so you can always re-process at a later date as technology moves on and everything is easy to find.
So for me at least PL7 is staying. PL9 is sitting on the self and I suspect will gather dust. I have trialed ON1 and I cannot grasp the workflow. I have trialed Luminar Neo and found that the masking is very good, certainly better than PL9. For landscape and other purposes, it is pretty good and cheap. I re-worked some RAW from my last trip to NZ and the results were so much better than the originals. Now that could be down to me failing to get the best from Photolab but how did I manage to get better results from a piece of software that I have only just started using?
