- This topic has 9 replies, 3 voices, and was last updated 27 minutes ago by
.
-
Posts
-
Hello Nitro community,
I am new to Nitro and this is my very first post.To start I would like to express my appreciation for the software, Nitro is nice and looks very well done.
For the time being I am a little bit struggling with Lens Correction. I went through the tutorial video and it looks quite easy but then, when I try to apply the same strategy to other pictures I realise it is much more difficult. What strategy do you use? Do you correct sample pictures in other ways and then use them as comparison to Nitro?
Thank you!
GaetanoIf you take a sample image and follow the instructions, then it should create the necessary information to automatically correct images. It’s hard to tell based on your description what is not working. If you aren’t using the sample image approach (grid lines), then you can use a RAW+JPEG pair and try to match them, but that can be hard depending on how exact you want the correction. I’m hoping to add more lens profiles to the app in the future, but there is so much other stuff I have to do first…
Hello Nik,
thank you for your kind answer.Probably my main mistake is that I am trying to determine the lens correction starting from images that were not taken for this purpose. I need to find a suitable subject and then try to apply the procedure you describe in the video tutorial.
About lens profiles, I am sure you have already considered using Lensfun library. Could it be an option?
Thanks a lot for your work!
Hello Nik,
thank you for your kind answer.Probably my main mistake is that I am trying to determine the lens correction starting from images that were not taken for this purpose. I need to find a suitable subject and then try to apply the procedure you describe in the video tutorial.
About lens profiles, I am sure you have already considered using Lensfun library. Could it be an option?
Thanks a lot for your work!
The video description has an attachment for an image that you can download and use as a test target. The video also shows how to take images to use to train the lens corrector.
Yes, I’m looking into Lensfun.
Thank you, I will surely check the video again.
About Lensfun, it is a fairly well-populated database maintained by a community, maybe it can save some work to support new lenses provided that there is a way to update it in Nitro.
Just a tiny statement… years ago, forum discussions split between people who liked their lenses optically well-developed and refused software help and others who just cared about the end-result, wether by skillful engineering or skillful measuring and creating “lens-profiles”. In times of AI “optically well-developed” became a label of old fashioned people. What I always stumble about: I pay for a good lens and if necessary also for a lens profile made by the manufacturer. If that company takes their job seriously I don’t have to bother software-engineers to come up with better profiles, no?
Anyways, I really prefer the manufacturer profile if available, as I made several pretty quirky experiences with DxO (not recognizing the lens or confuse it with another type of the same FL and aperture, just different manufacturers) and Capture One. I think, if a manufacturer decides to cut shortways, he has to be responsible for an excellent lens profile – not the programmers of RAW converters.
I agree with you.
Unfortunately, regardless of the source of the lens profile, there always is an impact on the developers of the RAW converters. As far as I know there is no “standard” format for lens profiles, not even an agreement on which parameters should be provided and this makes the developers work much harder.
Community based projects have the advantage to gather the contribution of all the enthusiasts who would like their lens to be supported along with a stable and homogeneous programming interface. A good fallback solution in caso of missing manufacturer’s profiles.I agree as well as I have some concerns based on my experiences with DxO 6 years ago. DxO claimed to have the best lens profiles, hand-knitted and specifically designed for certain lens/camera combinations. That was the reason for several problems: DxO confused one lens with another because the manufacturers, especially within the L-Mount Alliance, simply have no valid, open standard for the use of EXIF-data. Also, Sigma and Lumix (I believe, Leica as well) use different specifications in metadata for the same lens – I counted up to 7 different entries for the same 500/5.6 lens.
The same? No. The Nikkor 500/5.6 PF always had the same entry, although mounted to 2 different Nikon bodies. The Sigma 500/5.6 Sports was on 5 bodies and had 6 entries in the overview of lenses. Here it was the use of one of two teleconverters altering the lens’ name but this doesn’t explain a difference of 53 images between the count of EXIF lens entries (2106 in total) and body entries for the same lens (2159). Maybe Capture One is also messing up the data a bit, I used it’s filter functions.
Because we see the name of the lens in the metadata of the image, we tend to believe there’s some kind of standard, right? No. There are as many standards as lens- and body manufacturers and I don’t blame any software programmer for errors in the app caused by the manufacturer chaos.And then there’s firmware and hardware updates: the same lens mounted on the same body can get 2 different entries after the body or the lens got a firmware update. All of a sudden the model name was altered as manufacturers are too stubborn to express the aperture values as they are meant: f/16 is not F16 or f-16, and it’s rather point- and clueless to describe a lens as i.e. Zeiss Planar 1.4/85. Dividing 1.4 by 85 = 01.01647… stubborn stupidity! 85 mm ÷ 1.4 = 60.714 mm and that’s the diameter of the diaphragm called aperture @ f/1.4. Longer focal lengths factor the same aperture number as a bigger diameter, shorter as a smaller. Planar 85 mm/1.4 would be the correct terminus. Apparently every marketing department sees this numbers as their playground, add some other characters and a chunk of glass, plastic and metal suddenly looks worth the price.
Sorry for the rant.
If manufacturers cannot agree on a naming pattern of the two most important values of a lens, imagine the confusion between owners and contributors of a project.
Now, if manufacturers already avoid an agreement about the use of metadata, how will the various members of the project deal with the problem? In repeatably the same way for each lens? Even the ones which don’t have their metadata onboard?Let me say that this is a great post! I have experienced, in the past, some of the troubles you describe and I fully agree that sometimes it looks like that the lens is chosen just by chance. Nonetheless, I think I had such problems with maybe 5% of the pictures and once you have noticed the wrong guess you can often fix it manually.
I really like the manual procedure for lens distortion in Nitro and I like even more the fact that the correction data are saved and used wisely, but I would also love to have the choice to manually choose one lens from a database and see the correction sliders set accordingly.
That’s the experience I was thinking about, something similar to “Use Levels” in “Auto Enhance”.
Probably it is not easy to code, I understand, but it would provide a good and quick solution for who just doesn’t have time, chance, skill or will to fine correct.
- You must be logged in to reply to this topic.