Will jpeg XR format emancipate us from RAW conversions at last?

[sperera]

will shooting JPEG-XR within a year or two as it starts to appear in cameras emancipate us from proprietary software to manange RAW files?
"Why? Because 100% of the information contained within a raw file can be encoded into a JPEG-XR file (no loss), and the manufacturer can decode the file in-camera, giving us the photographer the default colors and look intended by the manufacturer.
In this way you get the look the manufacturer intended without having to do any extra work. The camera delivers a fully useable (demosaic'ed) JPEG-XR file which may be 100% lossless for quality or lossy to save space (your choice). The JPEG-XR file, which is a "normal" image file will be transportable and useable just about anywhere a JPEG file is today.
There is one drawback to this approach, that only the pickiest of the picky will get held up on, and that's the fact that because your image gets demosaic'ed at the time it is captured, you will not be able to benefit from advancements in demosaicing technology as time moves forward. Today, if you open your raw file from 2000 in Lightroom 2.4, theoretically, it could look better than it would have had it been developed 9 years ago, because demosaicers are better today. In practice, though, this is not a significant concern."
...the above is quoted from Hasselbladinfo.com forum member Bradley Gibson

[Panopeeper]

The author does not understand the essence of working with raw data. It has nothing to do with the lossiness of the in-camera processed images; actually, the old JPEG format does allow for lossless encoding (for example the Canon CR2 raw data is stored in this foprmat).

Recording raw data and processing it out of camera gives the photographer the chance to make adjustments based on decisions the camera could not make, and it allows for using different software tools.

The finished image out of camera (no matter in which file format) is the digital polaroid.

[bradleygibson]

The author does not understand the essence of working with raw data. It has nothing to do with the lossiness of the in-camera processed images; actually, the old JPEG format does allow for lossless encoding (for example the Canon CR2 raw data is stored in this foprmat).

Blanket, unsubstantiated opinions about what others do or do not understand add little to the value of the conversation.

I stand by my statements.

-Brad

[bradleygibson]

<deleted>

[sperera]

Blanket, unsubstantiated opinions about what others do or do not understand add little to the value of the conversation.

I stand by my statements.

-Brad

I hope we can get some good debate with this Bradley my friend.....the jpeg XR format sounds great to me!!!!!!!!!!!!

[Panopeeper]

the jpeg XR format sounds great to me!!!!!!!!!!!!

The JPEG XR format is great. I guess it will take over the "presentation segment", i.e. what JPEG and TIFF is used for today. It could be used for encoding the raw data as well, lossily or losslessly, in smaller storage than the encoding methods are doing that today.

The topic of this thread is not if JPEG XR is better than JPEG. The topic is, if the concept of recording more or less raw data in camera and processing it off-camera remains superior to in-camera processing, or if higher quality recording of the in-camera processed images makes the raw data obsolate. The form of storing the data (encoding) is irrelevant in this question.

The question is, if negative film would become obsolate if the polaroid's quality were better.

[bradleygibson]

The JPEG XR format is great. I guess it will take over the "presentation segment", i.e. what JPEG and TIFF is used for today. It could be used for encoding the raw data as well, lossily or losslessly, in smaller storage than the encoding methods are doing that today.

The topic of this thread is not if JPEG XR is better than JPEG. The topic is, if the concept of recording more or less raw data in camera and processing it off-camera remains superior to in-camera processing, or if higher quality recording of the in-camera processed images makes the raw data obsolate. The form of storing the data (encoding) is irrelevant in this question.

The question is, if negative film would become obsolate if the polaroid's quality were better.

Personally, I see the topic of this thread slightly differently.  But for the moment, if the question is, "which gives superior results, in-camera or post capture processing", I am in complete agreement that the post-capture results will be at least equivalent but almost always superior, (provided users are willing to use the appropriate tool*) because of a number of factors including, computing horsepower, application sophistication and flexibility.

If we ask the question if the amount of difference in the finished file will be great enough that photographers will, in general, be willing to invest the extra time for these incremental improvements, my answer changes--working professionals, consumers and most amateurs are unlikely to see or see enough of a difference to care.

I believe that most will not be able to see the difference, or enough of a difference to warrant the traditional raw-camera-processing post capture workflow in a mature JPEG-XR world (years away, I am sure).  I foresee the majority of photographers moving on from mosaic'ed formats and not looking back, especially sensor resoloutions as high as they are and still climbing.

So yes, calling it a 'convenience workflow' or a 'presentation format' is all fine--I believe it will become the defacto standard format people use to store, edit and share their files.  There will remain people for whom mosaic'ed raw files are not obsolete.

The speculation on adoption is a just a prediction (ie. opinion, not fact) based on experience as I have seen it, nothing more.

-Brad

* using the appropriate tool properly for optimal image quality can also be a non-trivial task, if one is not technically experienced.

[sandymc]

The JPEG XR format is great.

A question, and this is genuine ignorance, no agenda - I thought the XR spec was getting a fairly cool reception from camera manufacturers, etc due to the Microsoft patent situation? And yes, I do know about the MS Microsoft Open Specification Promise, but that has limitations - e.g., no sub-licencing, etc.

Sandy

[Jonathan Wienke]

Blanket, unsubstantiated opinions about what others do or do not understand add little to the value of the conversation.

I stand by my statements.

-Brad

You should stand way behind them; you're obviously the one who "doesn't understand". The whole point of shooting RAW instead of any flavor of file format processed in-camera (whether JPEG, JPEG XR, TIFF, or whatever else may be out there) is to have the flexibility of choosing any set of processing parameters in post instead of committing to any set of parameters in-camera. For applications where in-camera processing is "good enough", standard JPEGs are generally "good enough" as well. If standard JPEGs aren't good enough, then it's pretty unlikely that in-camera processing is going to be deemed desirable, either, and RAW/DNG is going to be the file format of choice. JPEG XR doesn't add enough value to either scenario to make a compelling argument for its existence.

RAW conversion isn't a slavery to be emancipated from; it's the freedom to process one's images as one pleases with whatever tool one deems best for the job. JPEG XR doesn't offer any meaningful advantages over RAW/DNG for those who choose to do their own conversion, and those who prefer in-camera processing (most likely press shooters with deadlines or the less technically astute) are unlikely to appreciate any great benefit from the image quality of JPEG XR over standard JPEGs. And as already pointed out, JPEG 2000 does everything JPEG XR does. JPEG XR is a solution in search of a problem.

[Eric Myrvaagnes]

Jonathan put it much batter than I could. Many of us want the freedom and flexibility to make our own decisions about processing. JPEG XR sounds to me analogous to a situation that might have happened back in film days, if some manufacturer claimed to have come out with the "ideal" film for all purposes, making all other films obsolete.

[bradleygibson]

Sandy, that stuff not something I deal with.  In any major effort like this there are many players, challenges, etc.  I do not know how well or poorly JPEG-XR is being received.  I hope it is well, but I do not know.  It was ratified as an international standard by the JPEG committee earlier this year--that's all I know at this point.

Jonathan put it much batter than I could. Many of us want the freedom and flexibility to make our own decisions about processing. JPEG XR sounds to me analogous to a situation that might have happened back in film days, if some manufacturer claimed to have come out with the "ideal" film for all purposes, making all other films obsolete.

Of course, JPEG-XR allows this--your camera files come out of the camera with some look to them.  And I'm saying that I think most folks (but not all) will find this to be a good starting place for their files.  I don't believe editing one's files from this point will not produce a noticeably inferior result to a purist raw workflow.  Given the simplicity, I expect it'll be popular.  But there's no way for me to prove any of that, it's just an opinion.

Eric, Jonathan, Gabor, I will take Jonathan's advice and will stand "way behind my statement" because according to you, I "obviously" don't understand.  The thread is yours.

Take care, fellas,
-Brad

[Panopeeper]

I am really sorry to have choosen too strong words in my first post and thereby started hunting Bradley from this thread. However, the essence remains unchanged: JPEG XR is an encoding and compression method. It's deployment may be meaningful as a substitute for the customary JPEG, as a substitute for the lossless JPEG or for any of the many different encoding/compression methods used by the cameras when storing the raw data. The question of encoding/compressing has nothing to do with the decision "raw or presentation image".

Neither the size, nor the quality of the legacy JPEG images is the cause for or against creating in-camera presentation images. Think of following: the modern, high-pixel count DSLRs create "fine quality" JPEG images in the size 4-10 MB (and more, I guess). For example one of the Canon 7D's images posted at Imaging Resources is 8 MB large. When resaving it in PS, quality 11 yields 6.8 MB and quality 12 yields 9.7 MB. In other words, that quality is at least as high as it is reasonable, probably higher, though the bit depth is a serious restriction. The JPEG embedded in a Canon 7D or 5DMkII raw file is about 2.5 MB. So, it does not depend on either the quality or the storage size. JPEG XR will save space at even higher quality, particularly it allows for more than 8bit depth.

The opening post started out with the question if JPEG XR will cause the elimination of raw image recording. The issue is not, which method to use when recording the data but which data gets recorded. There is no quality, which makes a presentation image equivalent to a raw image.

[Jonathan Wienke]

Of course, JPEG-XR allows this--your camera files come out of the camera with some look to them.

A RAW image doesn't have any "look"; the out-of-camera "look" depends totally upon which RAW converter one chooses and what default settings are chosen. When shooting RAW, you have almost complete freedom to choose any "look" you like, or try out several different "looks" to see which one is the most appropriate for the image. This is why most knowledgeable digital shooters use RAW. You have complete flexibility in processing the image in any way that you choose, and are guaranteed the availability of 100% of the sensor's dynamic range and color gamut.

And I'm saying that I think most folks (but not all) will find this to be a good starting place for their files.

On the contrary, there are two main classes of shooters who regularly use in-camera processing: media photographers who need the convenience of camera-produced JPEGS to meet their deadlines, and the technically clueless who either aren't skilled enough to operate a RAW converter and get good results or don't even know what RAW is.

I don't believe editing one's files from this point will not produce a noticeably inferior result to a purist raw workflow.  Given the simplicity, I expect it'll be popular.  But there's no way for me to prove any of that, it's just an opinion.

Au contraire, the experience of most photographers here is that it's not too hard to do RAW conversions that are noticeably superior to images processed in-camera. And exactly how is editing a JPEG XR image any "simpler" than editing a standard JPEG? The only advantage XR has is higher bit depth. You still have the disadvantages of a pre-baked color profile and white balance setting, and the irreversible loss of DR and gamut from the in-camera processing. The higher bit depth may make a newspaper photographer's job a bit easier, but they aren't typically looking for the absolute best image quality--meeting their deadline is the most important thing. For them, standard JPEGS are already overkill with regard to overall image quality; a Canon 10D can usually capture a better image than a typical newspaper press can print. And JPEG XR isn't going to help the technically challenged folks much; if you're too clueless to operate a RAW converter well, you aren't likely to be savvy enough to materially improve your images just because they have a few more bits per pixel.

The reason I and most other serious digital photographers use RAW is for the flexibility it offers when processing the image. No set of RAW conversion options is appropriate for every image, regardless of whether the conversion takes place in the camera or later on in a PC. JPEG XR does not change that fact, and as a result offers little benefit over standard JPEGs. JPEG XR may be useful as a format for presenting or archiving processed images, but touting it as a replacement for RAW workflow is simply ignorant.

[bradleygibson]

I gave it some thought, and decided that rather than abandoning the thread, to respond to respectful discourse and to ignore the rest.

I am really sorry to have choosen too strong words in my first post and thereby started hunting Bradley from this thread.

Thank you Gabor, no harm done.

However, the essence remains unchanged: JPEG XR is an encoding and compression method. It's deployment may be meaningful as a substitute for the customary JPEG, as a substitute for the lossless JPEG or for any of the many different encoding/compression methods used by the cameras when storing the raw data. The question of encoding/compressing has nothing to do with the decision "raw or presentation image".

There seems to be a chasm here between the science and the practice.  Every file format, including raw, is an encoding and compression (or not) method.  So of course I must agree with your assertion that "JPEG XR is an encoding and compression method".

But in practice, people choose encoding and compression method (aka file format) based on their own personal weighting of factors such as quality vs. ease-of-use.

JPEG-XR promises to offer some new in-camera combinations (demosaic'ed floating point linear encoding, for example, with ICC presentation profile) which can allow the files to come with a default look set by the manufacturer, that the user can change in post without quality loss (modulo floating point precision), relative to a traditional mosaic'ed raw workflow, in post.

That's where I think we're disagreeing or misunderstanding one another.

[Jonathan Wienke]

JPEG-XR offers some new combinations (demosaic'ed linear encoding, for example, with ICC presentation profile) which can allow the files to come with a default look set by the manufacturer, that the user can change in post without quality loss (modulo floating point precision) relative to a traditional mosaic'ed raw workflow in post.

That's where I think we're disagreeing or misunderstanding one another.

This functionality of a default manufacturer "look" is already available from cameras that output DNG-format RAW files; the default conversion settings are set in-camera. The difference is that the manufacturer's "look" isn't baked in the way it is with JPEG XR. With DNG, you can change the white balance simply by altering the channel multipliers on the linear RAW data. With any camera-processed image format, you need to reverse-engineer the camera profile and tone curve to guesstimate the original linear values, tweak the channel multipliers, then re-apply the camera profile and tone curves. It's a much less exact process, and doesn't work as well as simply adjusting the RAW conversion white balance settings. You're replacing best practice with a kludgy approximation by using JPEG XR instead of DNG, and not gaining any real convenience benefit over DNG. What's the point?

[Panopeeper]

JPEG-XR promises to offer some new in-camera combinations (demosaic'ed floating point linear encoding, for example, with ICC presentation profile) which can allow the files to come with a default look set by the manufacturer, that the user can change in post without quality loss (modulo floating point precision), relative to a traditional mosaic'ed raw workflow, in post

1. Already the last but one DNG format version, 1.2, allows for storing a complex set of information additionally to the raw data,which can be used by the raw processor to achieve certain "look". (The set of these tags together are called "camera profile".) The camera could store such data in the raw file, while the raw processor might use or discard it, under the user's discretion.

Adobe have provided sound reasoning why not to use ICC.

2. The definition of "raw image data" is not firm. Think of Canon's sRaw1, sRaw2 and mRaw - Canon calls them "raw", although they are far from the raw data; they are semi-processed. There are some good reasons to stick to the non-demosaiced data. Example: the very first stage of out-of-camera noise reduction should take place before or integrated in the demosaicing process.

I don't see any point in creating semi-raw but not directly presentable images (linear is not presentable). Either raw, or presentation version (both could be stored in JPEG XR format). Canon's semi-raw formats are nothing but space-saving features; they become obsolate with the very large capacity and cheap memory cards (they are anyway disastrous for those converting them in DNG format for workflow reasons).

[ejmartin]

Perhaps a bit OT, but why didn't JPEG-2000 catch on?  My understanding is that the compression algorithms were substantially improved.  Is it just the inertia of the status quo?

[Ben Rubinstein]

You can shoot a TIFF in nikon cameras still can you not? That's what Bradley is pretty much suggesting except now it will have a new name. Just don't try to recover blown highlights, open up deep shadows, try and achieve the ability to decide on an output colour space without lossy remapping, decide on a different sharpening or noise reduction to that done in camera.

Yes a 16 bit TIFF has a huge amount of information but it's still baked in, you are CHANGING the file rather than deciding on a new way to interpret the original 1's and zero's of the RAW data. I don't see that this would be any different. Just try recovering detail in a partially burnt out wedding dress from a TIFF, even a 16 bit one.

[Guillermo Luijk]

Perhaps a bit OT, but why didn't JPEG-2000 catch on?  My understanding is that the compression algorithms were substantially improved.  Is it just the inertia of the status quo?

I think this is for the same reason as most people looking for music, at least in their first quick search, do it in Youtube which is a video site. It's low quality audio is still good enough to listen to any audio clip. If you want more quality then you'll search for the MP3, but most people won't even go to that step.

Present JPEG is good enough to show your images on the net and even to print them. Why should users and the industry make any effort to change what already works fine enough?

I guess JPEG XR doesn't support undemosaiced data. That reason alone is enough to forget about it for quality digital captures.

BR

[bradleygibson]

You can shoot a TIFF in nikon cameras still can you not? That's what Bradley is pretty much suggesting except now it will have a new name. Just don't try to recover blown highlights, open up deep shadows, try and achieve the ability to decide on an output colour space without lossy remapping, decide on a different sharpening or noise reduction to that done in camera.

Yes a 16 bit TIFF has a huge amount of information but it's still baked in, you are CHANGING the file rather than deciding on a new way to interpret the original 1's and zero's of the RAW data. I don't see that this would be any different. Just try recovering detail in a partially burnt out wedding dress from a TIFF, even a 16 bit one.

Ben, one of the differences between JPEG XR and 16-bit TIFF is the fact that JPEG-XR has a specification for dealing with high dynamic range floating point information.  What this means in English, is black is 0.0 and white is 1.0 (instead of 0 and 65535 for tradtional 16-bit TIFF).  It is possible to write numbers like 1.2 or -3.4 with JPEG XR (there's no standard way to do this with 16-bit TIFF).  This means that IF you choose to edit and rewrite your data, in floating point format, it is quite difficult to clip your information.

Of course, you are still free to do a parametric description of your edits, just the way Lightroom does with all edits today (including JPEG), to retain maximum image quality, but the downside here is the portability issues (app B can't really know what all of app A's parametric edits mean until they're all standardized).

Both approaches have tradeoffs, but it's not such a lopsided question with this.

-Brad