I had a weird lens problem in the past too (opened a thread, no one really saw a problem in this case).
When I had it new, my Sigma 35 (L-Mount) was not recognized because there was no profile. The lens name was correctly shown in PL and I had to use manual lens correction. So far so good.
Later DXO made a lens profile for a Leica L-Mount 35mm f2 lens. At some point PL now showed me Leica 35mm as lens type for the images taken with the Sigma lens (for old images that once showed the correct Sigma name and for new images).
Finally DXO made a lens profile for this Sigma 35mm lens and it now shows me the correct name and uses the correct profile.
There is definitely no problem with the lens IDs because in the end DXO is able to distinguis between the Leica and Sigma lens and get the correct profile. IMHO the only one to blame is the lens selection algorithm and not the lens makers.
Luckily it happened only once. For all the other cameras and lenses it worked so far.
After two full weeks DXO responded and asked me to download a new copy of PureRAW 2 from the link they sent. This is after that reinstall. The lens is still wrong. This is an amazingly powerful RAW converter that I have praised to students and fellow photogs. I agree, the correct lens is in the EXIF info. Just read it.
Between this, and not having side-by-side image comparison, no ability to set custom hotkeys, and the obtuse way to put the goddamn tool palette on top of the photo that youâre trying to work on, Iâm struggling to stay with DxO. Only 3 pros that I see are: better raw interpretation than other programs, the lens profiles WHEN and IF they work applied for sharpening and aberrations, and DeepPrime for noise reduction⌠other than that, I think for me Capture One or even ON1 making more sense [+ Affinity].
Really sad to reach this conclusion, I had such high hopes for DxO PL5, then I thought they would fix it for PL6⌠and here we are
PL doesnât simply provide a lens correction module, like most (all?) other software does ⌠it provides a correction module specific to the {body + lens} combination used to capture the image.
A lens is a lens, a sensor is a sensor. I donât know (literally, I never made comparisons within the same mount or the same sensor with various lenses) how far a single combination with sensor X and lens Y in one module would get different results than module for sensor X in camera W plus the (correct) module for lens Y in front of it.
And sensors canât be changed in cameras, except using a modular system like Phase One or Hasselblad.
Sure, the questions are âhow do I measure a sensorâs spectral sensitivity?â and âhow do I measure a lens without some sort of sensor behind?â, so it makes sense and is probably cheaper to only publish real life (contrary to test lab equipment) results. But as DxO only tests one copy of a lens and one copy of a certain body behind, this completely ignores production tolerances of both items. How many of us did send a lens with front- or backfocus to the manufacturerâs service division and got the answer âeverything within the tolerancesâ?
I fail to see this system as superior to a real modular system. It makes itâs users to feel special when a combination in his possession is a match - or should be one because as some of us experienced, lens IDs are not always hitting the target. I know, getting lens ID correctly has nothing to do with measuring that lenses flaws and set up counter-measures. But if the recoginition already is so much (and so often) off, how trustworthy are the profiles?
As a developer myself, Iâm really tired of people saying stuff like âdevs do not careâ or âlazy devs!â (that last one is especially common in the video games world). First off: this is attribution bias, itâs tiring and not useful, maybe cut it off?
Devs probably care, but they already have work to do, objectives for their week or the quarter, and spending half their time on this forum to work on what the loudest people here have asked for may not be their actual job.
Secondly, the generic answer for âwhy does this software have bugs and limitations I donât like?â is:
Because software developers have time and resources constraints.
And they have to make choices and pick what bugs they fix and what features they implement or improve; those choices may not align with your desires.
There are differences between companies of course, and some do a great job with limited time and resources, other do an okay job, and yet others are struggling to ship stable and improved software. And that factor, however it goes, is almost never a personality or attitude issue (âcare/donât careâ, âmotivated/lazyâ), itâs the result of past and present management decisions.
That being said, DxO as a company should probably have a stronger presence on this forum, and maybe a clearer way to track issues and to communicate with users about progress. Of course, that too runs into time and resources constraints. But Iâve seen companies, including tiny teams, do much better at that than DxO seems to be doing.
To answer more specific questions:
Why would DxO not recognize a lens or get it wrong?
Because EXIF metadata is not standardized, vendors of camera bodies and lenses may put whatever they want and change that randomly with firmware updates, end users may modify the metadata using software, etc.
So you get a very wide variety of data (see also: Cory Doctorowâs seminal essay on unreliable metadata). Your software has to sift through this metadata and make educated guesses, which can be correct or incorrect. And it takes a human to identify that a guess was correct or not, the software cannot autocorrect itself to be magically right all the time.
You can also optimize your code to limit false positives (returning âyeah I got it, the lens is Xâ but X is wrong), but then you increase the number of false negatives (returning âsorry I did not recognize this lensâ, but the lens was actually X and the software has a profile for X, so you lost an opportunity to apply a correct profile).
Why donât DxO have profiles for every lens?
Because their lens profiles are camera body + lens profiles, and require having access to that camera body and that lens (which costs money), and making a real human worker shoot a bunch of tests in a lab (which costs more money), before the resulting images can be analyzed by DxOâs dedicated software.
So weâre back at the limited time and resources thing.
And DxO seems to favor making camera profiles for camera brand manufacturers (Sony, Canon, Nikon, Fujifilm, Panasonic etc.) and other established lens manufacturers (Tamron, Sigma), but so far not really for the upstart Chinese cheap lens manufacturers (Yongnuo, Yi, Meike, TTArtisan, 7Artisons, Rokinon, Pergear, etc.). Probably because, having limited time and resources, they chose to
Why doesnât PhotoLab let me enter my lens manually?
Because they donât have a profile for that camera + lens combination in their profile database anyway, so thereâs nothing you could enter that would make sense. You could tell the software âhey itâs a TTArtisan 50mm f/1.2â and the software would be like âcool, that information is meaningless to me, so Iâll do strictly nothing with itâ.
In the scenario where they do have a profile for your configuration and their code that detects the camera body and lens gets it wrong, then maybe they could let users search in their profiles database to apply one manually. But that would only fix rare problems, and could lead more often to users applying an incorrect profile and complaining about bad results. Iâd err on the side of âprobably a footgunâ in this case, but who knows.
Why doesnât DxO allow its users to use an existing, ready made lens profile from the basic source of lens flaws: The manufacturer and designers off the lens. If the poor DxO devs have limited time resources, lack of manpower an what else of excuses you bring up: why not using the most obvious help there is, like any other RAW converter company? And come up with a home brew DxO profile when itâs ready?
To be clear, C1âs lens profiles are not better (and this by a huge margin) than DxOâs. As well as C1âs support is nbc better than the one of DxO (at least according to some examples I read here. Some of my requests and bug reports date from November 2022. Unsolved until today and C1 still requests log files. No more. Waste of time, the can have it, if they pay me as alpha or beta ester.
But until DxO profiles become better I can use at least a better profile than no profile at all. A that is the laziness, the lack of decision and the courage. Be it management or who else will become less and less important when users see âanother app offers me a better way to get good imagesâ.
Youâre defending devs, but I tell you: No craftsmen, no handymen ever could get legally away with the poor quality devs constantly produce. In one company more, in others less, but after watching so many bugs, so much poorly designed software I now want to see them in front of a court. Then they can explain the reasons why their software has to be such a miserable mess (this is in no way only targeted at DxO, itâs going through all companies and the usual suspects are already sacking their employees not by ten or hundreds but by thousands. Why?). But no, of course, poor devs are never reliable for anything⌠and can exclude that reliability in every of their long novels of EULA.
DXO creates profiles of lenses and bodies combined.
Manually choosing a lens with an unmatched body would not give PL the benefits that a matched pair does.
That said - when PL failed to detect a lens on a body which already have been measured and matched itâs a less positive side affect.
One must be delicate when comparing different solutions against each others.
Assumptions are not beneficial in any way.
DxOâs approach seems to be âeither we support it with our custom high quality process, or we donât support it at allâ. If they offered a weaker form of lens profiles based on vendor-provided information, like Capture One or Lightroom, then users could compare the two and say that DxO is not better than the alternatives.
Ultimately itâs a commercial choice. No idea if itâs the best move, or if theyâre shooting themselves in the foot by being purists.
You can listen to people with experience telling you that reality is more complex that youâd like it to be, or you can decide that a whole profession is a coterie of bad actors out to get you.
Subject is: There are a lot of professions in need for a âprofessional liability insuranceâ. I think, you are the one who only knows programming reality and takes it for granted that bridges donât collapse, car engines donât explode, ships do not sink, water is coming out of the tap and electricity is constantly available.
âComplexityâ is a super lame excuse - who created problems so complex that no one can oversee all of them? Software is in no way a naturally grown organism with a normal pace to adapt and correct flawed code. Itâs nothing else than a game park for people who like to see themselves in an important, relevant position, plus for people who demand constantly âfaster solutionsâ.
And Iâm a person with experience, a lot of user experience - I think I could also live a life without a constantly flowing stream of more mistakes, errors, failures, crashes.
Again: Craftsmen and building companies, machine manufacturers, food producers are legally reliable for severe mistakes and can be punished, no matter how complex the problems are - software devs are free to fail without solid consequences. Now, proof me wrong.
I donât think they really fixed the underlying problem. When I remember right, the leica lens profile (which was selected as âcorrectâ profile) came betweend PL4 and PL5 and the finally correct Sigma profile between PL5 and 6.
So I had deleted that wrong profile that was downloaded⌠but decided to experiment with it.
And it works. Lens Sharpness, Lens Deformation⌠all the things that supposedly only work if the right camera body and precisely right lens are detected, the planets align and DxO blesses us with the privilege of applying their very special Secret Sauce to our plebeian raw files⌠which makes me think itâs all marketing, and theyâre selling snake oil.
The 2 lenses here [real: Yongnuo 25/1.7 detected: Lumix 25/1.7] are VERY different.
In my very ignorant opinion, DxO PL6 should not be âcorrectingâ effectively for a completely different lens than the one it thinks itâs correcting for. And yet, it does.
YN 25/1.7:
7 aperture blades
7 elements in 6 groups (including 1x Ultra-Low ED, 2x Aspheric Elements)
Min focusing: 20cm
Magnification: .2x
PL 25/1.7:
7 rounded aperture blades
8 elements in 7 groups, including 1 UHR (Ultra High Refractive Index) and 2 aspherical elements
Min focusing: 25cm
Magnification: .14x
