If you were not a member of the PhotoLab 9 Beta testing team perhaps you shouldn’t make assumptions about the participants, the quality of their testing, any issues they faced, the feedback they provided to DxO, or DxO’s response to that feedback.
Mark
I am fully aware of all the various issues that many users are having with PL 9. Happily, I am not one of them. However, I guess you did not read, did not understand, or are purposely misinterpreting what I wrote. When it comes to PhotoLab Beta testing your comments betray your ignorance.
Mark
Looking at the right side of the GitHub repository, you’d see a mix of TypeScript and Rust.
In the JetBrains 2025 developer survey, TypeScript was the primary language of 22% of developers and is used by 42% of developers, so again it looks like you’re taking your case for a generality.
TypeScript was created at Microsoft in part by the Turbo Pascal creator, to “fix” issues they had with JavaScript such as the lack of static typing.
But basically if you know JavaScript, you don’t have to learn much to use TypeScript instead.
Fun fact, Visual Studio Code is also mainly (>95%) TypeScript.
Coincidentally, React and TypeScript are what power the new notification center, the taskbar, and other “slow” new versions of components of Windows 11: if you’re fine with their presence with W11, you should be fine with React for an application.
See above, he won’t have any issue finding contributors for the TS part. Rust has a bunch of active developers too, see the above developer survey.
React is primarily used in websites, so as most websites are already “hand-coded” I don’t think it will be an issue.
Quickly looking at the source code, he uses Tauri (which seems well-documented) for the cross-platform aspect.
If you’ve never encountered frameworks like React then I understand your point of view. But I’ve seen React before and what I see here is rather clean.
Looking at the commits I’d agree that some parts look AI-driven (like shaders), which reflects what’s written in the README.md (Gemini helped him with “research and implement concepts like the Menon demosaicing algorithm”), but most of the codebase looks human to me.
It’s just using paradigms popularized in the past decade and not ones from before.
With open source, you only need one person with enough free time and determination to help a project a lot.
On the opposite, you want to avoid the “too many cooks in the kitchen” effect.
Tauri uses the OS’s integrated web renderer, so if the result is ugly in MacOS and not in other OSes then it’s Safari’s fault.
I understand that beta testers are under NDA and cannot legally say more, but the implications of this paragraph are worrying for the future of the software!
This is also very worrying because it looks like the DxOMark website hasn’t been updated with recent lenses.
Go to the lens section, set the date range to “2023 to 2026”, and only two lenses were tested in the past 3 years… and none at all after Feb. 2023.
If the website is supposed to be a showcase for their services, that’s bad.
So, no proper code then, just scripts. Whatever happened to well structured Object-oriented analysis design?You know the kind if code that can be decoupled and refactored on a class by class basis?
23 years ago, I was called in as a consultant to a company whose requirement was "to create a UI for an app that was the front end for a Java data engine.
So, I started by specifying protocols that described the conversation between the Java engine and the UI. Then the Java team wrote code that supplied data that would satisfy that protocol and I wrote a small test app in Delphi, that the Java team could use prove that the engine was working correctly.
In the meantime, I led a small team of Delphi developers in the creation of the UI, that drew its data from that JNI protocol, using OO design and coding techniques that promoted code reuse by way of classes and inheritance.
A few months later, when the day came to connect the two parts together, the project manager expressed her anxiety as to how difficult and time consuming she was expecting that process to be. She nearly fell off her chair when I proceeded to change about a dozen lines of code to join the UI at the Java engine.
I see none of this kind of structured approach in this source code. It just seemed to be a collection of functionalities written in code units with no context. Good luck trying to refactor that when changes and new features need to be made.
Fun??!! That’s what I would call frightening and so typical of M$. I am so glad I left Windows behind many years ago.
Web rendering? Aaaarrgghh!!! I am so glad that DxO uses platform native UI components, so it looks right in both Windows and macOS.
That’s what I want in my software. Something that looks right but doesn’t deliver……….
Upvoted, but I think that porting the software to Linux for a small company like DXO would be too much of an effort. The only sensible thing could be to become more Wine-friendly, similarly to how Valve handles Steam on Linux. Nik Collection worked absolutely fine under Wine during Google times, last time I checked the DXO version (and DXO Photolab … maybe around version 5) I didn’t manage to have it working.
However, my feeling is that DXO is going the opposite direction and such a solution is even less likely nowadays. During an interwebs outage I was really really pissed off when I could not use Nik Collection because it failed online check (Photolab worked fine). After that episode I decided to end my status as a DXO fanboi and I won’t upgrade every version, but just when I feel like an added value. Anti cheat code is one of the main reason why Windows Games don’t work under Wine/Linux.
Rust is proper (compiled) code.
The core of the program you’re talking about is written in Rust, then the Tauri framework loads a web rendering engine from the OS that loads the transpiled TypeScript of the UI alongside the React components, while the performance-critical code written in Rust stays in compiled form.
Rust may not be an object-oriented language the same way C++/Java/PHP are, but it has different mechanisms (traits) for composing structure properties in a “composition over inheritance” manner.
A growing mass of programmers argue that composition of traits allow for a better decoupling than inheritance of class (example: if Square inherits from Rectangle then it needs an override for the resize method as you can’t resize a Square and a Rectangle the same way, whereas composition allows each shape to only import methods that are relevant to it).
According to this logic, when you visit websites they don’t “look right” either because they don’t reflect the UI language of the operating system.
“small” of around a hundred employees… also I completely missed the news, but a Linux version for Affinity is now “discussed seriously internally” at Canva: I clearly remember on the Affinity forums (one or two years ago) that people who asked for it back then received very similar pushback as what I’m seeing here. Interesting.
Gareth seems to be suggesting things based on his opinions rather than fact. In addition, the lack of current lens reviews on DxOMark’s website is irrelevant since it has been a completely separate company since the beginning of 2018 and does not create the lens profiles for DxO Labs.
DxOMark is now focused on mobile imaging devices rather than DSLRs, mirrorless cameras, and interchangeable lenses. which is evident from the reviews on their website. DxO Labs tests and creates their own lens profiles in-house.
Mark
Mark,
Assuming your reply represents the current situation, in which the company that licenses PL does its own measurements of lenses (and bodies, etc) to create the modules that are used in PL for corrections, then either the other company left all of the hardware and software required for module creation with the PL company, or the hardware and software was replicated at the PL company. Did DxO Labs (PL) duplicate all of the DxO mark infrastructure?
Both DxO Labs and DxOMark were originally a single company under the DxO Labs brand. DxO Labs was founded in 2003. The first version of Optics Pro, the predecessor of PhotoLab, was release in 2004. DxOMark started off as a image quality team at DxO Labs. Around 2008, that image quality team created a separate benchmarking website owned by DxO Labs and called DxOMark. Fast forward to the beginning of 2018 when the DxOMark team parted company with DxO Labs to form a completely separate company called DxOMark. This coincided with DxO Labs releasing the first rebadged version of Optics Pro as PhotoLab 1. As an aside, PhotoLab 1 was the first version of PhotoLab to have local adjustments.
DxOMark, as a separate company, no longer focuses on DSLRs, mirrorless cameras, or interchangeable lens, and has not done so for at least the last two to three years. That company’s current consumer area of interest is in mobile imaging devices, e.g. smartphones and portable computers. Additionally, as a separate company, DxOMark has never created lens profiles for the software published by DxO Labs. I hope this clears up some confusion regarding the historical relationship between DxO Labs and DxOMark before and after they became separate companies.
Mark
Now that is what I would call decidedly weird. I have been teaching and using OO design patterns since before the famous GoF Design Patterns book was published in 1994. One of those patterns is Composition, which has absolutely no need for inheritance. Instead, classes can be made to comply with multiple protocols and delegation. Swift uses extensions to concatenate behaviours without the need for inheritance.
Very poor example. A square is simply an equilateral rectangle and can be resized using exactly the same code. No need for any different code.
I do not see a direct answer to my question: the hardware and software required to properly evaluate a lens and a digital camera, and then to build a good module to correct the lens optical compromises during manufacture (but not to do this for every sample of a lens as there are manufacturing tolerances and some samples that somehow passed QC are “substandard”) is not insignificant. The simple modules of FastRawViewer (used by Topaz, as one example) do not produce the same results as what I have found from current PL Elite Complete – the “corrections” are better with current PL. (Aside, because of manufacturing tolerances, does DxO evaluate multiple samples of a lens?) When DxO split into two firms, did the PL firm keep the testing and evaluation hardware and software or was the hardware and software duplicated?
It’s a secret, shhhhh.
If DxO Mark Kept the data, then they would license it……………… Has anyone noticed how loads of companies have lens correction now? And many don’t have to tie the lens to the camera. You simply get the lens correction that applies to the lens, regardless of the camera used.
I wonder where they bought all that data?
One of the open systems lens correction persons asks a user to take specific images of a target and then transmit those images, from which a module may be made. The results in my opinion are not consistent because some of the submitted images are not taken correctly.
I do not understand being a “secret” as to what was the disposition of the lens measurement and thus correction hardware and software to make a PL lens module when DxO became two separate and independent firms.
I guess the secret is…… who kept the data?
Once again you are talking out the opposite end from your mouth. As I pointed out at length and in detail in this thread a few days ago DxOMark does not now, and has never had, anything to do with the creation of PhotoLab’s lens profiles. They continue to be created in house as they always have been. I assume you will attempt to reinterpret what I posted on the subject to support your agenda. So be it. I expect that from you. But regardless of your whether or not you are willing to accept them, facts are facts. For the sake of your edification I am repeating below what I posted earlier.
And, by the way, the reason that DxO’s lens profiles are so highly regarded is precisely because they are generally superior to the profiles of other companies due to their greater precision. The price for that is a longer wait time until they are ready. Perhaps lens profiles and image quality are not that important to you as I recall you were recently extolling the quality of ON1. As with most of PhotoLab’s competition I have owned licenses to every ON1 version since 2018 and I can say with confidence that it continues to be very inferior to most of the high end post processing software available today including PhotoLab.
Mark
I think its can be pretty simple this days - may they just read once from the lens the embedded (distortion, vignetting, CA) correction info - and not measure this things. May easier to measure just the lens - may its even don’t need a camera.
Regarding color is simple, just measure on a few chart. Or may far easier - may based on with just few jpg it’s can be done.
And may few company just measure / sell data to everyone - Think about like company LensRentals has more than enough lens and cameras and expertise to do that. May even few people company can do it (in some quality) - i think the main problem to get the equipment’s.
I remember (?) something like for DxO its takes approx 5 day to do one profile.
I processed a photo with nothing more than lens correction with Luminar Neo / On1 / DxO - really couldn’t see much difference between any of them.
I ran my X-H2S + EF100-400 Mk2 through Luminar and On1 and n either of them barely made any change at all.
Maybe other lens need more help, who knows but for me, too much maybe being made of lens corrections?
Roll on to 2026 and it’s clearly not about price any more. Even banks and wealthy corporations are ditching Windows for Linux. Personally, I have come to so despise Windows that I am considering alternatives to PL and CorelDraw if I can’t take them with me to Linux.
How many times do folks have to say, it ain’t never going to happen for DxO.
As for disappointment, you made your first mistake when you chose Windows.
It took me a few years but I finally fell for Apple and have never been disappointed since. Just think about it, Linux is a hobbyist’s version of UNIX, macOS is a heavyweight, secure, industrial version of UNIX with loads of free, competent, easy to use apps included that I have never had to pay for List of built-in macOS apps - Wikipedia.
For photography…
Since I don’t have either the time or the inclination to faff around configuring Linux or complaining about what isn’t available for it, I work every day with my safe, secure, Mac, for which I don’t even need anti-virus.
As for quitting DxO, just because they are not prepared to accommodate an OS that only really lends itself to servers and nerds, no thanks, I have serious photography to do.