Talk:Raw image format

I merged the "CCD-RAW" and "RAW image file" articles to create this one, and added some new text that corrects some errors in the previous versions (particularly in RAW image file). It can still use a lot of work; I don't know when I'll get a chance, so if someone else wants to contribute please do.

The color depth bullet is probably still confusing. Raw files have 12 bits per pixel, but only one channel (each pixel is either red, green, or blue depending on what color in the Bayer filter covered it. JPEG files have 8 bits per pixel for each channel (red, green, blue), so a total of 24 bits per pixel. But it's difficult to compare color depth between a 1 channel and a 3 channel image. I don't know if this needs to be explained here, in the color depth article, or not. If someone can explain this better, I'd appreciate it. If this is really confusing, let me know and I'll work on it.

A couple of new topics that might be useful:

• When NOT to use raw; a lot of people are very vocal about it being a waste of time: "shoot correctly and you don't need it"
• A list of common conversion programs (e.g., Capture-One, BreezeBrowser, Bibble, Photoshop CS and Elements 3)
• A list of file extensions for common raw formats (e.g., Canon uses CRW, Nikon uses NEF) (the German version has an incomplete list for starters)
• Some raw formats include an embedded JPEG, I think for quick extraction to evaluate overall quality before making the effort for a quality raw conversion; I don't know whether this is universal or why it is really used so didn't include it
• More than just a pointer to Adobe DNG, although this could wait a few years to see how well it is received

--Rick Sidwell 20:45, 10 Dec 2004 (UTC)

"Raw files have 12 bits per pixel, but only one channel (each pixel is either red, green, or blue depending on what color in the Bayer filter covered it. JPEG files have 8 bits per pixel for each channel (red, green, blue), so a total of 24 bits per pixel."

That is interesting I would like to see that in the article, about how each pixel is a certain color due to the physical sensor having R, G, and B sensors.

I will try to add mosaicing to the article. However the quality of JPEG vrs Raw is simply untrue. JPEG is not RGB. Each RAW pixel contains 12 bits of brightness information. Whereas JPEG includes only 8 bits of brightness information. Color and saturation are (depending on the parameters) stored at considerably lower resolution. Therefore a RAW image will always have sharper brightness details than a jpeg even if they were both from a better external source. This is a crude simplification of JPEG btw. In reality it is more of a mosaic itself with colors and brightness gradients defined at intervals. --Darkfred ^{Talk to me} 21:48, 17 July 2006 (UTC)

Marketting aspects

While RAW support is a standard feature on dSLRs, it was also included in a lot of compact cameras several years ago, but few people used that because memory cards were expensive. Now memory cards are cheap,but it seams marketing has found RAW to be a "pro" feature and they use it to differentiate product categories. Example: The Nikon Coolpix P5100 and P5000 cameras are quite high quality products but they don't have RAW support while their predessor the Nikon Coolpix 5000 included it. —Preceding unsigned comment added by 80.133.144.123 (talk) 16:05, 14 February 2008 (UTC)

Conversion

Can raw files be losslessly converted to PNG? TIFF? dbenbenn | talk 15:09, 20 July 2005 (UTC)

I don't think the conversion would be reversable, and a certain amount of information would be lost.

You can to a perfect bit-for-bit lossless conversion to Adobe DNG, which is an unusual type of TIFF. With PNG and normal TIFF, you only lose from round-off error and from the loss of little odds and ends like knowing the alignment of the pixels to the camera sensor. It's not like going to a GIF or a low-quality JPEG. --anonymous

But the conversion is still not reversable from .DNG, simply because the digital development settings (contrast, color, curves) are not replicated 1:1 between adobe raw and the various manufacturers conversion utilities. There might be other information changes, I don't know if .dng supports the full range of bit depths that various manufacturers have. .DNG attempts to be a simple unified format whereas the manufacturers each have specific quirks in what data is saved and how it is stored. --Darkfred ^{Talk to me} 15:13, 13 April 2006 (UTC)

Oh no. Optionally, Adobe DNG can cheat. It can store all the unrecognized and unconvertable parts of the original file as a great big arbitrary blob. When you convert back, the original can be reconstructed bit-for-bit. AlbertCahalan 04:49, 16 April 2006 (UTC)

(However, overexposed areas are just as white with 12 bits as with 8, so using raw is not a substitute for correct exposure.)

This is not entirely true. Most blown out areas are only blown out in one or two channels. Adobe's latest version of camera raw software can recover a brightness value from these channels, that is quite realistic, and mix this with the color values from nearby pixels to produce nearly perfect recovery of some highlights. (i have used it to recover entire faces correctly). --Darkfred ^{Talk to me} 11:50, 22 September 2005 (UTC)

Accuracy

I have tried to make the article more accurate by explaining the mosaicing process and changing the comparison with JPEG to relate to how JPEG is actually stored. (previous editor was confusing JPEG with uncompressed RGB, and confusing mosaiced 12bit RAW with 36bit RGB). My grammer is not the best, feel free to make it more readable. --Darkfred ^{Talk to me} 22:24, 17 July 2006 (UTC)

List of cameras supporting a RAW format

This section must be in it's own article, if it should exist at all. When complete, it will be too long for the RAW article. For example, all cameras listed on thedcraw-page should be included in this list. Berland 13:29, 18 February 2007 (UTC)

How about convert to list of raw formats, with corresponding companies, instead (like the box at top but with more info maybe)? Dicklyon 15:16, 9 March 2007 (UTC)

That is much more interesting than a long list of camera models Berland 16:21, 9 March 2007 (UTC)

I started the list of camera models with raw, and a similar section in TIFF. Uninteresting I agree, but I was looking for a camera with raw mode and couldn't find a comparison table anywhere, and ended up making a list myself. It's the sort of useful information that a continuously updated encyclopaedia excels at. So I think that this list belongs somewhere on Wikipedia, though not necessarily in the article on raw formats. Pol098 02:24, 10 March 2007 (UTC)

That's what this is for. Dicklyon 06:13, 10 March 2007 (UTC)

Thanks for that, I'd never have created the list if I'd known that site. Though I still think the list is worth having, perhaps on its own page. Pol098 23:47, 10 March 2007 (UTC)

Are there any objections to moving it to its own page? Berland 10:15, 10 March 2007 (UTC)

Noone seemed to oppose this, so I did the move, see List of cameras supporting a raw format. Berland 16:59, 17 March 2007 (UTC)

Questionable

However, RAW permits much greater control than JPEG for several reasons:

Finer control is easier for the settings when a mouse and keyboard are available to set them. For example, the white point can be set to any value, not just discrete values like "daylight" or "incandescent".\

The settings can be previewed and tweaked to obtain the best quality image or desired effect. (With in-camera processing, the values must be set before the exposure). This is especially pertinent to the white balance setting since color casts can be difficult to correct after the conversion to RGB is done.

• Both can be done on JPEG images using appropriate software. Similarly they can only be done using appropriate RAW software. I understand there are apparent quality differences between doing this on a 'altered' image and a 'unprocessed' image, but the article currently seems to suggest it isn't possible in JPEG. ny156uk 20:32, 2 April 2007 (UTC)

Your point is correct. The important point in this context is only the final quality, which is limited by either 8-bit for JPEG or 12/14/16 bit for RAW format. --Berland 20:47, 2 April 2007 (UTC)

What's not possible with a JPEG is any recovering of information in pixels that have already been clipped to the colorspace gamut. It's not because it's JPEG (same deal with TIFF), but because it has been rendered to a colorspace. With raw, there is no such clipping gamut, just sensor saturation limits. But you're right, the article does need work. Dicklyon 23:35, 2 April 2007 (UTC)

Capitalisation consistency

The article uses "raw", "Raw", and "RAW" inconsistently. Someone who knows more about what is acceptable should advise on which one to use throughout the article, and use just that one. 221.29.158.74 00:08, 6 May 2007 (UTC)
(Erk just not logged in)

Per comment below, lower-case is more logical. Anyone object if I change it? Dicklyon 23:09, 15 May 2007 (UTC)

I don't call it RAW because—in the words of the article—I believe it is a single file type, but simply because it is a file type. The lower-case spelling is also logical. --Adoniscik (talk) 00:58, 22 February 2008 (UTC)

Raw is not a format

Raw is just a collective term for the different camera manufacturers properitary file formats for saving unprocessed images directly from the CCD. Therefore it should be written raw and not RAW --EivindF 14:51, 8 May 2007 (UTC)

I agree, logically. But about half the books that talk about RAW use all caps, as if it was a file type. I think that's because conceptually it is a type of file, even though maybe different actual types and formats. Dicklyon 15:08, 8 May 2007 (UTC)

Images

I suspect that these images could be used in this article, but I don't know enough about them to be sure, perhaps editors might find them useful?:

• Image:Minolta Dynax 7D RAW Camera White Balance.jpg
• Image:Minolta Dynax 7D RAW 6000K.jpg
• Image:Minolta Dynax 7D RAW 3500K.jpg

Cheers--DO11.10 22:37, 15 May 2007 (UTC)

Someone already added a set like that in the color balance article, which seems more sensible, since they have more to do with white point and color balance than with raw format. Although, if he doesn't fix the license, those will disappear. Yours are a bit unclear, as the Exif says Auto White Balance on all of them. How were these images made? Dicklyon 23:03, 15 May 2007 (UTC)

I'm sorry, I plead ignorance. I am just the image fairy.--DO11.10 00:48, 16 May 2007 (UTC)

Requested move

The following discussion is an archived discussion of the proposal. Please do not modify it. Subsequent comments should be made in a new section on the talk page. No further edits should be made to this section.

RAW image format → Raw image format — More typical and logical case; stop propagating misconception that it's an acronym —Dicklyon 19:35, 27 May 2007 (UTC)

Survey

Feel free to state your position on the renaming proposal by beginning a new line in this section with *'''Support''' or *'''Oppose''', then sign your comment with ~~~~. Since polling is not a substitute for discussion, please explain your reasons, taking into account Wikipedia's naming conventions.

• Support (I'm the nominator). Even though half of books and other sources use it as "RAW", there is no logical basis for that all-caps. We seem to have a consensus, and a consistent usage in the article, to go with the the typical and more logical generic word "raw", so we ought to move the article to reflect that. Dicklyon 19:40, 27 May 2007 (UTC)
• Oppose (somewhat weak oppose) - Most papers I've seen use the all caps, too, and I agree at least half the books use this. Perhaps the reason is because they want to more clearly indicate it is in fact a file... and while it might not be a standardised format, it is indeed a type of file that has many commonalities between the different incarnations of it (size, what data is there, etc.) Even though it might be better if the world didn't use it all in caps, since it kind of does I would tend to oppose the move here. I believe it is more likely someone would search for RAW than raw, and there is no reason why this is inherently ungrammatical or incorrect. (Feel free to present evidence, especially about common usage, to the contrary.) --Cheers, Komdori 17:56, 30 May 2007 (UTC)

The evidence is very mixed, which is why I figure we get to choose. GBS shows about 2/3 of books use all caps in "RAW file format", but that drops to less than half in phrases like "raw conversion". Uwe Steinmueller, who wrote one such, mixes both forms on his web pages. Adobe uses "Camera Raw" without all caps; same with Extensis in their "Pro Photo Raw Image Filter"; but Microsoft and OpenRAW went the other way. We should certainly acknoledge both in the lead, but should pick one to use consistently in the article, and make the title match it. I chose one, not realizing I wasn't going to be able to do the move, due to some history. If you'd rather go with all caps, will you do the work to change it all back that way? Dicklyon 04:21, 31 May 2007 (UTC)

• Oppose: See [1], [2], & [3]. These are all major voices in technology & they use "RAW". (Wikimachine 02:39, 4 June 2007 (UTC))

So true. Yet other respected major players like Adobe and NIST and authors Bruce Fraser, Stephen Johnson, and Nick Sullivan don't all-cap it. So why side with Microsoft? Dicklyon 03:15, 4 June 2007 (UTC)

• Support - It isn't an acronym, but rather the English word raw. I see no reason why it should be capitalized. However, I think it would be useful to add a sentence or two to the (newly moved) article stating that although it is not an acronym, in common usage people often type it as "RAW". Josh Thompson 08:04, 4 June 2007 (UTC)
• Support - same reason as Josh Thompson above. --Berland 14:53, 4 June 2007 (UTC)

Discussion

Any additional comments:

Does anyone understand why the article and the talk page have different names and cross-redirects? Did someone do an improper move at some point? Dicklyon 03:17, 4 June 2007 (UTC)

Ah, yes, here's the diff. Some anon IP changed all the "raw" to "RAW", and then in the next diff changed the article to a redirect to an article of the current name. Too bad it wasn't noticed and repaired much earlier. Now we have a mess. Dicklyon 03:21, 4 June 2007 (UTC)

Oh, and before that, it has been moved from RAW to raw, properly. Like I'm proposing now. Dicklyon 03:25, 4 June 2007 (UTC)

The above discussion is preserved as an archive of the proposal. Please do not modify it. Subsequent comments should be made in a new section on this talk page. No further edits should be made to this section.

It was requested that this article be renamed but there was no consensus for it be moved. But, since the split history had to be merged anyway, I have ended up moving the article to be at raw image format, if only because that did away with the need for moving this talk page. --Stemonitis 09:16, 6 June 2007 (UTC)

Thanks; with only one weak oppose and one Microsoft supporter in opposition, the consensus seemed close enough. Thanks for fixing the previous improper move. Dicklyon 14:36, 6 June 2007 (UTC)

Apple Aperture raw processing limited to selected cameras.

"Finally, in October, Apple released Aperture, a photo post-production software package for professionals whose chief feature

is full support for raw files."
 Actually it only supports a few cameras' raw images as of 7/22/2007. See the post on the apple site below.

http://discussions.apple.com/thread.jspa?threadID=828344&tstart=105

I was quite suprised to find my Panasonic DMC - FZ8 raw files imported with a blank "Unsupported Image Format" message.

LoghouseJD 19:04, 22 July 2007 (UTC)loghouseJD

Proposed merge from digital negative

Since the guy who tagged it didn't start the discussion, I will.

Support – the two articles seem to be on the same topic. Adobe's "Digital Negative Format" is a specific raw format, but that's not what the digital negative is about. Dicklyon 02:32, 15 August 2007 (UTC)

I'm the guy who tagged it. It seemed to me that it's really the same concept: "digital negative" is slightly more abstract, and this page is more concrete, but that's it: all discussion about why it's good to have a digital negative for an image, what you can do with it, who supports etc, that's one article. Stevage 07:20, 15 August 2007 (UTC)

Support - there is not enough common knowledge on the abstract "digital negative" to warrant its own article yet. --Berland 20:21, 8 October 2007 (UTC)

I did a merge. Feel free to work it over. Dicklyon 06:19, 22 October 2007 (UTC)

Inaccurate/Misleading Information

Camera raw files have 12 or 14 bits of intensity information, not the gamma-compressed 8 bits typically stored in processed TIFF and JPEG files; since the data are not yet rendered and clipped to a color space gamut, more precision may be available in highlights, shadows, and saturated colors.

It should read:

Camera raw files have 12 or 14 bits of intensity information, not the gamma-compressed 8 bits typically stored in processed JPEG files; since the data are not yet rendered and clipped to a color space gamut, more precision may be available in highlights, shadows, and saturated colors.

TIFF files can contain 16 bits of information per channel, so it is not a good example. —Preceding unsigned comment added by Xenoptrix (talk • contribs) 06:02, 22 October 2007 (UTC)

It's not really inaccurate, since the 8-bit processed images are indeed typically stored in TIFF files, and since TIFF files typically store 8-bit images. Yes they can also store 16-bit images, but that doesn't make it untrue or misleading. Dicklyon 06:06, 22 October 2007 (UTC)

Well a TIFF file can have 8 bit content, to say raw's benefit over tiff is that tiff can store only 8 bits per channel is misleading, as tiffs can store either 8 bit per channel OR 16 bit channel information. Thus, I would say remove TIFF from the discussion, because this section is really talking about RAW versus JPEG, JPEG can ONLY be 8 bits per channel.--Xenoptrix 06:14, 22 October 2007 (UTC)

Yes, that would be misleading if one said that. However, it doesn't say that's a benefit over tiff, it says it's a benefit over 8-bit images. The rest of the statement "since the data are not yet rendered and clipped to a color space gamut, more precision may be available in highlights, shadows, and saturated colors" should probably be clarified as being an advantage over tiff, however. Work on that; taking TIFF out is going the wrong direction. Dicklyon 06:41, 22 October 2007 (UTC)

"it says it's a benefit over 8-bit images" however, TIFF is implicated as being an 8 bit per channel format. I think this comparison is rather lax, it needs a more rigorous comparison. Because 8/16 per channel of the TIFF or 8 per channel JPEG image is NOT the same 12/14 bit per channel of the raw image. Instead the TIFF/JPEG channels are reinterpretations of mosaic pattern. Upshot: the term bit-depth has been applied in two different contexts, but for each context, bit-depth means something different.--Xenoptrix 07:32, 22 October 2007 (UTC)

How do I send a message to you, Dicklyon? I am new to wiki editing, it seems like you only see my comments here after I've edited the wiki entry. --Xenoptrix 06:26, 22 October 2007 (UTC)

All the pages I edit are on my watch list, so if you address me on a talk page I'll see it; or use my page User talk:dicklyon. Or use the email link on my user page to send me an email. Dicklyon 06:41, 22 October 2007 (UTC)

Innov8or's large edit

Innov8or (talk · contribs) recently checked in a very large edit. I'm highly suspicious of such large edits, but Google doesn't turn up any sources for copyright violation. Going through the material he changed, I notice he changed the release year of Apple iPhoto 5 from 2005 to 2006. A Google search turned up the Apple press release announcing iLife '05 (which included iPhoto 5). It was indeed released in 2005 - a year before LightZone, which Innov8or has been advertising in this and several other articles. Given his history (and obvious motivations) and the tone of the new material, I'm inclined to just revert it. However I'm wondering if any of it is actually useful and could be rescued with a lot of editing. --Imroy (talk) 10:35, 18 December 2007 (UTC)

Yeah, I agree... questionable although no obvious copyvio. Beyond that, his material was unsourced. I think it was appropriate to rv. Parts of it might be useful if sourced. ǝɹʎℲxoɯ (contrib) 16:33, 18 December 2007 (UTC)

I reverted it before I noticed the comments here. It's a big opinion essay mostly. Probably has some good points, but without a source to back them up, they seem to be more about his agenda as you note. Innov8or, feel free to contribute, but preferably in manageable chunks that make verifiable points with an encyclopedic style. Dicklyon (talk) 16:36, 18 December 2007 (UTC)

TIFF is NOT a RAW file... or is it?

The table "RAW image file" at the top of this article incorrectly lists TIFF (.tif) as a RAW file format for Kodak. Clearly, this standard exchange format is not a RAW file. However, the TIFF format is technically a general-purpose container format that could theoretically contain any arbitrary data (including RAW subpixel data) if a few engineers were sufficiently out of their freaking minds and applied for a private range of tag values for that purpose....

So the question is this: did Kodak really do something as bonkers as writing RAW subpixel data into a TIFF file, or did somebody just throw up a list of formats that Kodak cameras support and not check it carefully? If Kodak did do something so utterly bizarre, this should be clearly noted. If not, the mention of .tif should be removed.

Either way, the mere thought of embedding RAW subpixel sensor data in a TIFF container is going to give me nightmares tonight. If I wake up crying, I'm blaming all of you.  :-D

P.S. You left out the Rollei .rdc format, the Sinar .sti format, and probably a bunch of others. See http://pa.photoshelter.com/help/tour/formats for a large list and search for "raw".

Dgatwood (talk) 07:33, 28 January 2008 (UTC)

Raw images necessarily come from a camera or scanner?

Is it necessarily true that a raw image must come from a camera or scanner? It is my opinion that a raw file is a raw file regardless of its source.

I believe that years ago (I think before scanners and digital cameras were common), various sorts of image files regarded as "raw" existed; e.g. a screenshot taken directly from the framebuffer would be regarded as raw. There are old programs such as rawtoppm (part of the netpbm software), which considers RGB data to be raw. (Confusingly, netpbm's ppm format exists in "raw" and "plain" varieties, which are different.)

Anyway, my point is that the term "raw" for image files has a long history and hasn't always referred exclusively to digital camera or scanner images. 71.82.211.210 (talk) 22:46, 23 February 2008 (UTC)