Skip to main content
Participating Frequently
February 17, 2023
Open for Voting

P: - Add ISO and Exposure Time to the Noise Profile Tag implementation

  • February 17, 2023
  • 24 replies
  • 1856 views

Hello everyone,

 

I have a question (or a remark) about the "noise profile" tag that is expected to be found on .dng files.

For reminding, the noise in a raw picture can be approached by a Gaussian noise, whose variance scales linearly with the brightness. The noise profile tag therefore gives the slope and the intercept of this linear relation, for each color channel.

 

There are several methods to estimate these parameters, but I doubt that such an estimation is done everytime I take a raw picture with my smartphone. I suspect that the noise profile is estimated one time by the camera manufacter, and scaled according to the picture settings.

 

It is now fairly known that the slope scales linearly witht the ISO, and the intercept scales quadratically with the ISO, and such relationships can indeed be observed on the metadatas of pictures taken with different ISOs. 

Note that there is a first gap in Adobe's documentation, since there is no convention on the ISO for which the noise model is given. Many people refer to noise profile as the slope and intercept for ISO 100, whereas it seems in practice that the noise profile found in the EXIF is given for the ISO of the picture. Is this left to the control of camera producers, or is it controled by the dng requirement ? 

 

My second remark is about the exposure time. The intercept of the noise model is expected to scale linearly with the exposure time, and this can indeed be observed by manually estimating noise on raw images. However, the noise profile tag seems to be purely ISO dependant, and does not change slightly with the exposure, which is completely misleading.

 

This lead me to suspect that some camera manufacters obtain the noise profile tag by simply scaling a predefined noise profile using the ISO, WITHOUT considering the exposure time. Not only it is misleading, since the noise profile in incorrect, but the resulting tag cannot be manually scaled since the reference exposure of the noise profile is unknown !

 

This is why I believe the following points should be adressed in the dng specifications:

- For which ISO is the noise profile tag given

- For which exposure time is the noise profile given

 

I am aware that providing an accurate noise profile is hard, especially because many factors such as the temperature can have an impact. However, exposure time can vary a lot and significantly modify the intercept, and is very easily measurable.

 

I would be glad to hear yout thoughts on the matter, if you have any further information, please let me know.

24 replies

TheDigitalDog
Inspiring
February 24, 2023

Both of you don't seem to understand what is a noise profile. The last two answers are factually wrong :

 

That may be so. But the burden of having a feature request taken seriously is all on you, and so far, it is getting zero traction and zero votes. So I'll move on after repeating what I suggested earlier: let's sit back and see what all this back and forth gets you. As someone that has spent three decades working directly with Adobe on requesting features and beta testing, I'm in violent agreement with this opinion thus far (and over time, we'll see the reality): "I am sure of is that you are tilting at windmills."

Adiós

Author “Color Management for Photographers" & "Photoshop CC Color Management/pluralsight"
Legend
February 24, 2023

The actual RAW data is changed by in-camera noise reduction. There was a dustup because Sony and Canon both have cameras that apply NR and in some cases it can't be turned off. On cameras where NR can be toggled, your RAW file would be different. That right there negates a single noise profile.

And one thing I am SURE of is that computers will continue to get faster and that processing individual images will become the norm. Lightroom already does this with CA, they correct it based on analyzing the image and not from a lens correction database.

The other thing I am sure of is that you are tilting at windmills. I'll repeat my assertion that this idea won't have any tangible benefits.

Participating Frequently
February 24, 2023

Both of you don't seem to understand what is a noise profile. The last two answers are factually wrong :

 

"These tags that you want don't even make allowance for whether the camera applied noise reduction."

Noise reduction is a digital process applied during raw processing pipeline, whereas the noise profile describe the noise of the raw unproceessed image. It is not related at all to camera denoising or whatever, because the raw data is untouched.

 

"I expect future iterations of noise reduction to use AI and handle analyzing/profiling on an individual image basis."

 

Camera AI noise reduction algoritmh make use of the noise profile... They do not rehandle profiling for each image because it is too computationaly costly, and the precision of estimation can vary a lot depending on the image. 

 

You don't seem to acknowledge that per-camera noise profile is a thing, yet it is globally admitted that it is working and widely used by denoising softwares and raw processing themselves. Lightroom uses this data too if I am not wrong. I am simply suggesting to require a supplementary tag, that will add even more precision to the model.

TheDigitalDog
Inspiring
February 24, 2023

I expect future iterations of noise reduction to use AI and handle analyzing/profiling on an individual image basis.

 

Indeed! AI Noise Reduction in the raw pipeline, one that treats DNGs or proprietary raws equally, will render (no pun intended) this request moot. 

Author “Color Management for Photographers" & "Photoshop CC Color Management/pluralsight"
Legend
February 24, 2023

I understand that you are passionate abut this idea but sorry, not something that is going to produce any gains in image quality, easy of processing, or performance. These tags that you want don't even make allowance for whether the camera applied noise reduction.

I expect future iterations of noise reduction to use AI and handle analyzing/profiling on an individual image basis.

TheDigitalDog
Inspiring
February 23, 2023

Problem?

I guess we'll see how many votes get generated and what problems get, a.... fixed. 

Author “Color Management for Photographers" & "Photoshop CC Color Management/pluralsight"
Participating Frequently
February 23, 2023

Most smarphone output dng format. If they have a private tag as you say, I have, as a user, no way of knowing which tag it is. I may ask Samsung what tag they are using, but I will have to ask Apple too, and then Google, etc, because the tag will be different everytime. It would simply make more sense to introduce a new public tag. It is very likely that, given that there is no public tag associated with the noise profile exposure, manufacters will not bother to ask Adobe.

 

This is the problem; how is one supposed to read the noise profile tag in the current situation? My point is that the public tags do not allow for an accurate reading of the noise profile, because manufacters are not asked to give an exposure value. Even if a value was given, it would be under a private tag that would need to be fetched for every camera manufacter which is a real pain.

TheDigitalDog
Inspiring
February 23, 2023

You can count on one hand the number of cameras that natively save a DNG but good for them.

If they have private tags they wish to apply in a DNG, they can do so.

Tell me what camera manufacturer that writes DNGs has an issue with the tags and isn't getting assistance from Adobe? What cameras are you manufacturing?

Is this a solution in search of a problem? 

Can't upvote without knowing; sorry.

Author “Color Management for Photographers" & "Photoshop CC Color Management/pluralsight"
Participating Frequently
February 23, 2023

Let's say that I am an honest camera manufacter, that pays a lot of attention to tag accuracy. I do not want to use a proprietary file format, and I want to output raw file using dng format because it is widely used. All metadatas are filled without problem, and I finally get to this "noise profile" tag. I read the .dng specification to know exactly what to put there. I am then stuck because it is not mentionned which exposure I should consider when estimating the noise model !

If I chose an arbitrary exposure, I cannot mention it anywhere because the dng format does not support such a tag. 

 

By "adding a tag", I mean that Adobe should update the dng standard, add a new box and say "manufacters are expected to put the noise profile exposure there". Nothing more. The responsability of filling correctly the metadatas goes to the manufacters, but the responsability of asking them relevant informations goes to the file standart and conventions.

TheDigitalDog
Inspiring
February 23, 2023

The information probably comes from the manufacter (I am not 100% sure). Adobe are in charge of the meaning and instructions of the tags, and as of today the noise profile tag will always contain contain junk, independently of the good will of manufactors, because this tag alone simply has no meaning

 

It is not a possibility for me to upvote since I still have no idea what you expect Adobe to do. I suspect this will be the case with others and certainly Adobe. 

 

There is some data; you're not sure where it comes from, and again, I have no evidence Adobe is making it up and placing it into the DNG. The data has "no meaning" but that's a problem from the source; you can't tell us where this comes from. IF a camera manufacturer is placing bogus (or as bad, proprietary metadata) into a raw before it's converted to DNG, how is this a DNG issue? And what about the raw prior to conversion to DNG? 

 

If Adobe where to add a "noise model exposure" tag, the noise profile could actually make sense, and manufactors would be to blame if there was junk in the tags.

 

It only makes sense if the data is accurate; where do you expect Adobe to get this data? 

 

My remark concerns dng instead of tiff, because dng is a widely used format for smartphone raw.

 

Makes zero difference! DNG, like TIFF is a container. If you get garbage data and embedded into either, it's still garbage (or if you prefer, ambiguous). Both TIFF and DNG are Adobe's openly documented file formats. But anyone can embed allowed data, useful or not, into them. How is this an Adobe issue if some software (or camera) embeds bogus data into their container and what are they supposed to do, remove it? 

Author “Color Management for Photographers" & "Photoshop CC Color Management/pluralsight"