I fully understand the proprietary and secret nature of the DxO mathematics and algorithms, rather like the trade secrets behind the best Cremona violins, trade secrets – once these are lost – that are very difficult to regain. However, the issue remains the same: if DxO must test each body with each individual lens (to be correct, there need to be multiple samples of each to obtain a representative population of manufacturer allowed tolerances that also are not published), the number of “optical bench”, etc., tests that need to be conducted are rather large. However, if any supported body module can be used with any supported lens that the body will accept, again, using whatever proprietary “mathematics” and programs that DxO does use, then the number of total tests is much much smaller (a matter of combinatorics). The latter is my guess and more or less substantiated by your comments. Thus, once the Z9 is a supported body, and any supported lenses that the Z9 will accept in either Z mount or with a FTZ in F mount (but not necessarily individually tested, lens by lens, on a Z9 – not feasible for a short time interval “loaned” Z9), such combinations should thus be supported in DeepPRIME or the successors available. From my direct experience, I would agree with your statement about the noise reduction (and image enhancement) of DeepPRIME, although the Topaz current “AI” products are getting closer (and possibly superior on very specific individual images, but not in general).
I don’t know anything about their testing procedures and I’m fairly certain they are not about to divulge any of that information to the public. For me, what is important is the end result rather than the process they use to get there. Surmising how they test might be an interesting intellectual exercise but it holds very little interest for me.
Mark
It seems you misunderstood the intent of my question – one that can be answered by DxO staff such as Svetlana G. If each camera body must be tested with each possible lens than can fit it and that is supported by DxO (e.g., for the Z9, all Nikon Z mount lenses, plus using a FTZ II, all Nikon, Sigma, Tamron, … , F mount lenses), then the number of tests that DxO must do becomes very large (“elementary” combinatorics). If, on the other hand, DxO has a method to combine camera body “modules” with lens “modules” to produce an overall supported configuration for DeepPRIME, then the combinatoric issue does not arise. (Eg, mathematical convolution as the provably correct approach – although DxO will not comment as to what DxO does, there are only a limited number of mathematical methods that work based upon proven theorems – and if DxO has a new proven theorem, it is unfortunate that such an intellectual advance in mathematics has not been published in any of a number of mathematics/applied mathematics academic research journals.) Svetlana, any comments without revealing DxO intellectual property?
I will be very surprised if she gives you any feedback at all on the process, but I’ve been wrong before.
Mark
I am not asking about the internals of the “process”. I am asking that once DxO has a Nikon Z9 for “testing”, the module thereby developed will allow DeepPRIME and the image detail sliders (that require a supported body plus combination to be allowed to be used) to work for that body plus all the lenses that DxO supports that will fit that body, either Z mount or FTZII plus F mount. There are a large number of F mount lenses, both from Nikon and from other manufacturers for which DxO appears to have a module.
would seem to indicate that DxO uses each lens and each body together (but, as expected, gives no indication how these tests are then used for camera and lens support “modules” in workflow such as PL-Elite). If this is correct, the number of combinations is very large, and DxO would continue to need obsolete bodies (e.g., Nikon D300) run through all of the new lenses that the body could support, as well as new bodies such as the Nikon Z9 combined with a huge number of lenses, some of which are obsolete but still in service with some photographers. Thus, I still suspect that the lenses and the bodies have separate modules. DxO does advertise the extensive database of camera and lens tests that it has and thus can create the necessary modules for PL, but this does not mean that every camera and lens combination individually has been tested for PL support.
I’m not aware that DxOMark Image Labs is creating the body/lens optics modules for DxO Labs, the publisher of PhotoLab, They have been two separate companies since 2017. Perhaps their services are used by DxO Labs, perhaps not. It isn’t clear, and the subject of a processional relationship between the two firms hasn’t been discussed here to the best of my knowledge.
Mark
Thank you for that clarification – was the URL given by a previous respondent not in fact from the DxO web site? If so, then indeed the DxO Mark, etc, measurements indeed may not be upon which the DxO PL modules are based. In any event, this still does not address a question: when a new body that can accept many, many lenses already in the PL data base (my example is the Nikon Z9) – once the new body is added for PL, will the existing PL lenses that the body can use now be added so that DeepPRIME as well as the sharpening sliders (e…g., microcontrast) actually work for the body/lens combination? I am not asking about the algorithms, mathematics, internal measurements that generate a body or lens module for PL, but whether once a new body is added that accepts lenses that already exist or are added to PL “modules”, will PL support both the body and lens combinations? E.g., the Z9 plus FTZ will accept a Sigma 105 2.8 current F mount macro and that lens is supported (last I checked) – will the Z9 plus that lens be “fully supported” with DeepPRIME?
DxOmark has been separated from DxO Labs for a few years indeed. Nevertheless, I’d be astonished if the two companies would leave each other in the rain, “Labs” reinventing and redoing tests while “Mark” does these tests just for the fun of being able to present lens rankings…
Anyway, whatever the DxO siblings do and how or why they do it, I have to take what comes off them, no matter what my or anybody else’s bright ideas might be…
Hello,
we look to cameras and lenses announcements and request a sample of camera or lens to manufacturer.
The thing is we need a commercial sample to be sure we provide profiles (cameras or lenses) for average products which are available only very shortly before the commercial launch.
But the real issue is we are dependant of when manufacturer can provide us the samples. Sometimes we can get them before launch to market but the most of the time it’s after.
When we get the samples we can do shooting, measurements, profiles and validation (which we do ourselves, DxOMark is NOT involved as we are different companies). And when it’s ready, at last, support of the cameras in our software depend on release dates which are decided not only on camera support avaibility.
Regards,
Marie
Marie, Thank you for the explicit clarification, but there is one small matter that was not clear. Again, I am NOT asking for the proprietary algorithms, mathematics, etc., used by DxO. HOWEVER, to use DeepPRIME and the other full controls/sliders that PL"N"Elite has, the body and lens must be supported. Once a body (eg, Z9) has a profile for PL, etc, and a lens that can be used with that body (in the case of a Z9, including the FTZII that makes no changes to the optical or sensor characteristics of the Z9) also has a profile, will that body plus lens now be allowed to use the full features of PL Elite, or only HQ, not DeepPRIME? (NOTE: there are a very large number of F mount lenses from Nikon, Sigma, Tamron, etc, that have DxO PL profiles available and that will operate with a Z9 FTZII – will all of these now have full DeepPRIME?) A small second point to clarify what your fellow DxO staff person stated: the Z9 clearly is being introduced during the current production release interval of PL5E. In the event that the profile for the Z9 is not released by DxO until PL6E, as the body was in production during the PL5E epoch, will that profile be allowed to be used by DxO in PL5E, or will one have to license the update to PL6E? (As you may surmise from the above questions, I routinely use DeepPRIME with my presets for routine processing from NEF to clients – this approach automates much of my workflow.) (As for the samples from Nikon, as DxO is located in the EU, I assume these samples come from Nikon EU, not Nikon headquarters in Japan.)
Hello @wildlifephoto ,
Nikon Z9 will be supported during PL5 lifetime, don’t worry. But I can’t know yet when it will be supported.
When we support it you will have modules for all supported compatible lenses. For Nikon Z cameras we have 250 lenses supported at the moment, and we will increase it.
And there is no relationship between optic modules and functionnalities. Only Fuji X-Trans don’t have Prime (but they have DeepPrime) and that’s due to sensor type.
About RAW compression, if they are true compression (RAW S isn’t) we will do the work to support them.
Regards,
Marie
Marie,
Thank you for that reassuring clarification. At the moment, in terms of compression, I can provide a public URL:Unanswered Z9 questions from which I excerpt:
What are the file sizes? 55MB Lossless raw, 33MB High Efficiency Raw Star, 22MB High Efficiency Raw, 24MB JPEG Large Fine. End excerpt.
Obvious, the JPEG file formats are demosaiced and thus not usable with the current approach of DeepPRIME. There may be a still larger full raw NEF, along with the lossless raw, etc, of the above list. Will all of these “raw” formats be supported, and in PL5E or will one need to wait until PL6E? As we all know, Nikon (and other manufacturers) provide firmware updates that may change/add-to the above choices.
One small point about lens profiles. My current primary body is a Nikon D850 (soon to be a Z9 if the body does everything all of the reviews to date indicate that it will). As I cannot handhold a 600 4, and as my Sigma 150-600 Sport was stolen (along with the rest of my field kit due to the issues in the San Francisco Bay Area for professional photo/video gear by organised professional criminal theft entities), I switched to a Sigma 60-600 Sport. I also use a Sigma TC-1401 (1.4x) when I need a larger image of a subject (typically, followed by a crop in PL-E). For the 150-600 plus TC-1401, PL-E had no issues and recognized what I was using. However, the 60-600 plus TC-1401 is NOT recognized and I do NOT get DeepPRIME but only HQ (forcing me to use Topaz Sharpen AI that is better than anything I have seen from Adobe but not quite DeepPRIME). As I know that the TC-1401 is recognized with the 150-600 S and the 60-600 S is recognized, can you please get the Sigma 60-600 Sport plus TC-1401 recognized? I strongly suspect that I shall be using this same lens with a Z9 FTZII body. I have put the request into the DxO channel, but this should be very minor as DxO has profiles for all of the components and simply somehow ignores the combination.
Hello,
It’s not normal you can’t have DeepPrime on your photos.
Could you, please, provide some RAW samples from the 60-600 plus TC-1401 ?
Please, upload them via upload.dxo.com and let me know when ready.
Best regards,
Marie
This was my error – DeepPRIME does work, but the Lens Sharpness sliders are missing. I use these under many circumstances. The Sigma 60-600 Sport plus Sigma TC-1401 is seen in the EXIF (using an EXIF reader application) of an affected image as:
FocalLength: 850.0 mm
Lens: 60-600mm f/4.5-6.3
LensIDNumber: 191
LensID: Unknown (BF 38 56 A6 34 40 4B 4E)
LensSpec: 60-600mm f/4.5-6.3 G VR [6]
Note that the FocalLength is clearly longer than the 600 mm of the lens indicating that a TC was used. I assume that the hexadecimal digit code for the LensID can be used by DxO to identify the actual lens.
I will supply an image.
Marie,
I just uploaded one of my copyrighted NEF files (not a keeper) that has the issue. If lens sharpness had worked, this might have been a keeper as the bird (cropped) is the subject, not the whole scene. The NEF is from the camera, no adjustments. I am not adding regards, etc, to this message because this is not email and true names are not used here unless the Forum name chosen was the true name. You should have my real email from the upload. Thank you again for your answers.
Can you confirm that the optics module for that camera, lens, and TC combination was downloaded? if it was, there may be a corruption of some sort. I ran into a similar problem where lens sharpness was not showing up for my Nikon Z fc NEF files. Consider removing the camera/lens module from within PhotoLab and download it again…
Mark
Is there a mechanism to identify a lens (including TC) combination when processing an image to force PL to use the correct profile/module? If not, when I am loading an image into PL, it seems to automatically detect the lens from the EXIF and then downloads the modules it needs. How does one manually remove a module and then force a reload of a module?