17

Question about RAW advantages over JPG made me curious if somebody has examples where in-camera JPEG is actually image quality-wise better than RAW image converted in computer (possibly by third party RAW converter). I don't mean default settings, but the best you can get from both.

EDIT: I finally found at least one example myself: http://theonlinephotographer.typepad.com/the_online_photographer/2010/06/iso-6400-from-an-ep1.html

Although this is really subjective, I get consistently better colors from Canon's DPP (which should match the camera algorithms) than what I get from the converters I've tried. This might fall into (poor) skill category though.

EDIT2: Another case where this could possibly happen is when the highlight rescuing functionality (Active D-lighting/Highlight tone priority/...) is used. So if anybody has made this kind of tests, feel free to share your results.

EDIT3: Here are my own results where in-camera noise reduction seems to beat everything else: Does "long exposure noise reduction" option make any difference when shooting RAW?

Michael C
  • 175,039
  • 10
  • 209
  • 561
Karel
  • 10,729
  • 6
  • 48
  • 78
  • By better, are you referring to Image Quality? As @decasteljau has pointed out, there are some advantages Jpeg's inherently have over RAW, but I suspect you are asking about IQ. – Alan Aug 19 '10 at 20:23
  • Sorry, that's correct, clarified the question. – Karel Aug 19 '10 at 20:36
  • Interresting method in that article, however it should be possible to reach the same result just from the RAW file. I wouldn't mind having a go at it... – Guffa Aug 19 '10 at 22:18
  • Once I started using Topaz Denoise, I haven't looked back. It might be interesting, however, to try that trick after aggressive noise reduction to see if it brings even more of the detail back, despite the Topaz plugin being superb at it already. – Joanne C Aug 20 '10 at 01:06
  • 5
    An interesting thing about the article linked, is that the Camera JPEG is still just processed RAW data. The author of the article stated he knew of no way to clean up his noisy ISO 6400 RAW, however I don't doubt for a moment that I could get just as good of results with Lightroom and the RAW as the in-camera JPEG. One way or another, both are still just processing red, green, and blue pixels from a bayer matrix...if it can be done with JPEG, it can be done with RAW. – jrista Aug 20 '10 at 20:14
  • For EDIT2, see here: http://photo.stackexchange.com/questions/2005/how-does-adaptive-dynamic-range-work/2007#2007 – eruditass Aug 20 '10 at 20:27
  • @Eruditass - in general, I know how it works. I'm curious what happens if you put camera JPG side by side with computer-processed RAW, where the JPG is optimized for highlights (D-lightning/Highlight priority) in camera and RAW uses all the headroom available. – Karel Aug 20 '10 at 20:40
  • 1
    @jrista - "if it can be done with JPEG, it can be done with RAW" - you're making one assumption here that can't really be checked: RAW contains all the information available to camera when processing to JPG. Going further - the camera can even alter it's behaviour to get better data in the first place when shooting JPG (ie underexpose for highlights and apply correcting tone curve). – Karel Aug 20 '10 at 20:50
  • 2
    @matt, Ctein may know a lot of tricks, but that doesn't mean he keeps up to date with noise reduction, an area that is constantly changing. He used NoiseWare, which has been overshadowed by LR3 and Topaz DeNoise. Using LR3, there's no way one would still have that kind of chroma noise. Here is a shot from my K-x at ISO 102400: http://imgur.com/ejmX4.jpg

    Chroma noise can be removed fairly easily now. Of course, you can overdo it and have color bleeding, but Ctein is nowhere near that at ISO 6400, even on a m43.

     – eruditass Aug 21 '10 at 01:29
  • 2
    @Karel, JPEG by definition cannot have any more headroom than RAW. How it looks depends on how you apply the tone curve, but there cannot be any more detail in the JPEG in the highlights. As stated in the link, enabling highlight priority will cause the RAW to be underexposed.

    Now, if you want to compare a RAW without highlight priority and a JPEG with highlight priority, that's a different story. Shutter/ISO/Aperture will be completely different and I wouldn't consider a meaningful comparison.

     – eruditass Aug 21 '10 at 01:36
  • @Eruditass - I don't agree. If the only setting I change on my camera is from RAW to JPG and then get possibly better results (I'm asking if they are better, not assuming), it's meaningful comparison. BTW, shutter speed and aperture are left unchanged, so it's not a different picture. – Karel Aug 21 '10 at 07:37
  • 1
    But 2 exposures are 2 different pictures, even with exactly the same settings... essentially if you have a RAW+JPG of the same exposure, it will always be possible (not necessarily easy) to convert the RAW into at least as good quality using post-processing. If the exposures are not identical, then all bets are off. – chills42 Aug 21 '10 at 14:44
  • 1
    @Karel, if the only thing you change is RAW to JPEG, the highlight priority setting is not changed, you are in the first situation I described, not the second one. The JPEG cannot be better. Changing highlight priority will change at least one of the 3 exposure settings. – eruditass Aug 21 '10 at 17:20
  • 2
    I don't understand you, you don't understand me. Everybody keeps saying "by definition". I'm not interested in theory. I'm interested in the practical corner cases where shooting JPEG has an edge, whatever the reason was. I shoot mostly RAW myself, sometimes both and I'm really not into religious pro-raw arguments. – Karel Aug 21 '10 at 19:17
  • 2
    @Karel, if the only thing you change is RAW to JPEG, JPEG will never have an advantage. If you switch highlight priority as well as RAW to JPEG, it is a different story, because you are changing the exposure. In all cases, where the only variable is the RAW/JPEG setting, JPEG will never have an advantage. Is that clear enough?

    FWIW, I shoot JPEG over RAW on several occasions for other reasons.

     – eruditass Aug 21 '10 at 20:16
  • 1
    Usually, enabling something like highlight tone priority does not allow the camera to "better use" sensor data. All it really does is change exposure settings behind the scenes, which is not making "better use of the sensor", its really just doing what you can do yourself with raw, without actually telling you thats what its doing. Like Eruditass has stated, in an apples-to-apples comparison, shooting JPEG will never result in a "better" shot than a manually processed RAW. This idea about a camera theoretically being able to to better processing because its "in camera" is just that...theory. – jrista Aug 22 '10 at 02:47
  • 1
    I've read a lot over the last few days on several forums, including the one linked to Ctein's article. I see a whole lot of speculation and contemplation about how the in-camera algorithms can be specifically tuned for the particular hardware quirks in every single individual camera. While it is an interesting concept, from a realistic, practical, mass-manufacturing standpoint, I don't think it is practical. Its not a viable business decision to spend time meticulously tuning a JPEG algorithm to every piece of hardware it runs on...that would increase cost and diminish ROI. – jrista Aug 22 '10 at 02:49
  • I am beginning to wonder if this topic has gotten off track a little? Or perhaps it is just too subjective a question? I'm not really sure we can get a solid, objective discussion going, as there seems to be a lot of opinion and subjectivity revolving around this particular topic. I had to end some of my own comments a couple days ago as they were headed in a undesirable direction. To offer an alternative venue, it might be best to continue discussion on this topic, which I do find interesting, in the Photo.SE chat, rather than in comments. http://chat.meta.stackoverflow.com/rooms/154/photo – jrista Aug 22 '10 at 03:01
  • @jrista - I agree that it's gotten off-track here so I'll leave it as is. My intention is to always get the most out of my camera and as I don't change them so often, I've time to learn it's tricks. This train of thought has lead me to at least one important discovery (regarding noise reduction). – Karel Aug 22 '10 at 08:05
  • @karel — previous long comments discussion aside, could you edit the question to separate your own answer from the question itself? See http://meta.photo.stackexchange.com/questions/1601/should-edits-which-add-the-accepted-answer-to-the-question-be-reverted – mattdm Dec 23 '11 at 16:31

11 Answers11

12

No, by definition there isn't.

In the conversion to JPEG a lot of information is thrown away. The RAW file contains all the original information, so anything that the camera does to convert it to JPEG can be done later on from the RAW file. Generally you can even get a better result, because you can fine tune the conversion depending on the picture, and also because the RAW conversion program isn't contrained by the speed requirements that is put on the conversion in the camera.

So, if you get a better result from the JPEG file, it's simply because you don't know how to get the same result from the RAW file.

Guffa
  • 21,309
  • 3
  • 58
  • 85
  • No is probably the correct answer 99,9% of the time, but I'm wondering what the 0,1% is:) See my edit. – Karel Aug 19 '10 at 20:38
  • 1
    There is no 0.1%! Every JPEG produced in camera comes from raw sensor data. To the best of my knowledge, any individual sensor specific tuning adjustments are applied to the raw data at the analog-to-digital conversion step, not at the digital raw to digital JPEG conversion. So the advantage of any individual camera tuning is available both to the in-camera JPEG and the post processed raw file. – Michael C Jan 05 '16 at 07:21
11

This is a difficult question to answer, because yes it's technically possible to have a in-camera jpeg look better than RAW, but the scenarios are contrived.

Scenario One: Person Doing the Raw Conversion is unskilled

In this case, the user who is interacting with the RAW file doesn't understand what they are doing, and makes setting adjustments that produce an image that looks worse than the incamera jpeg.

Scenario Two: The raw converter itself is poorly written

Raw conversion requires an algorithm to reconstruct the image data from the bayer pattern data. By definition, the bayer pattern has 33% of the information that was present in the original scene. It's conceivable that one could write their own RAW converter using a poorly thought-out conversion algorithm which would result in a malformed image that looks considerably worse that what In-Camera processing can produce.

Scenario Three: The raw converter does not properly understand the RAW format

If you attempt to use an older RAW converter on RAW files from an unsupported camera, the results are undefined and likely will be terrible.

Alan
  • 18,717
  • 1
  • 69
  • 97
  • 2
    more scenarios: the camera knows something the converter doesn't (recognizes the lens and corrects for it), the camera does something the converter couldn't (underexposes for highlights). – Karel Aug 20 '10 at 19:38
  • @Karel: I think this idea about the camera having a finely-tuned algorithm specific to its hardware is intriguing, and a great theoretical discussion. However, I think it is something that is becoming unduly prolific in the online photographic community, and it really needs some concrete evidence to support it before we start assuming that its true. Evidence along the lines of an official manufacturer specification that clearly states they fine-tune their jpeg algorithm to each camera, or add knowledge about lenses. From a manufacturing/cost perspective, I don't think that its realistic. – jrista Aug 22 '10 at 15:59
  • 1
    @Karel - that is scenario 2/3 – Steve Ives Jun 30 '15 at 12:59
  • 1
    Any lens knowledge applied to the raw data in-camera can also be applied to the raw data in post-processing out of camera. If it can't in a specific instance, it is due to the specific raw converter's lack of such capability as written (scenario 2) and not to a limitation of processing raw files out of camera as opposed to in camera. – Michael C Jan 05 '16 at 07:28
4

Here some advantages of JPEG over RAW:

  • Faster to write file to card
  • Faster to transfer files from card to computer
  • Faster to browse files
  • Faster to process files in software
  • Less space on card/disk/backup
  • Universal support

Those are advantages that needs to be taken in consideration.

If you do NOT plan to retouch your images (take them as-is), JPEG could be the way to go. Because of the high-dynamics in the RAW files, the initial look may look a bit more flat on RAW, compared to JPEG. When converting to JPEG, the camera applies a light contrast curve on the image. Depending on your camera, JPEG may look better out-of-box.

decasteljau
  • 853
  • 1
  • 8
  • 13
  • 1
    You got me wrong, I mean pictures. Your answer goes better with this question: http://photo.stackexchange.com/questions/15/what-are-the-pros-and-cons-when-shooting-in-raw-vs-jpeg – Karel Aug 19 '10 at 20:34
  • My mistake leaving it too vague in the question, sorry. Made the clarification afterwards. – Karel Aug 19 '10 at 20:40
  • My last paragraph may be covering your question a little bit. – decasteljau Aug 20 '10 at 01:40
2

If the camera maker has special knowledge of the sensor characteristics, for example the exact absorbtion spectrum of the filters, it might be able to do a better job than a generic converter.

This is only theoretical, I don't have any good examples.

Certainly if you're going to work with JPEG anyway, having the camera do it will be faster than going through an extra software step.

Mark Ransom
  • 1,495
  • 10
  • 12
  • But the camera maker can write an external raw converter that uses the same exact knowledge of sensor characteristics as the converter they write into the camera's firmware. There is no difference, in terms of the computed result, between doing math to digital data in camera and doing the exact same math to the exact same digital data outside the camera. The lower processing power available and the need of doing it as fast as possible when done in-camera means in many cases an out of camera algorithm can be more thorough and precise than an in-camera one. – Michael C Jan 05 '16 at 07:36
  • @MichaelClark I don't disagree, but you're arguing a point that's opposite to the question. "Third party software" is even mentioned in the title. – Mark Ransom Jan 05 '16 at 14:56
  • "I don't mean default settings, but the best you can get from both." This kind of points away from your interpretation of the question and, to my reading of the question, makes it more about whether there is an inherent disadvantage to raw conversion out-of-camera vs. in-camera. Either the in-camera conversion algorithm or the out-of-camera conversion algorithm can be better or worse than the other. There's nothing that inherently gives the in-camera conversion an advantage, but (battery) power consumption requirements and time restraints do limit in-camera in a way the other isn't. – Michael C Jan 07 '16 at 11:47
2

I've always shot with just jpeg. I know many think I'm crazy but for me I find the advantages of jpeg (as mentioned by decasteljau) still just clinch it.

On Saturday, I set my Canon 7D to record both jpeg and RAW for every shot.

Below are the results after separately editing each of the file formats from the same shot in Lightroom 2.6

This is not really a useful test and you shouldn't draw any conclusions from it since I haven't edited each one in the same way. To be honest, any differences you see here are almost certainly due to the way I edited them rather than the format. The jpeg shot is much brighter for a start. If I get a chance, I'll have a 2nd attempt at trying to get the photos looking more similar.

It is however what the question asked, and I was personally curious to see if the difference was worth it.

Final edit from camera-generated RAW source image: alt text

Final edit from camera-generated jpeg source image: alt text

matt burns
  • 6,036
  • 4
  • 39
  • 52
  • It takes a lot of practice and experience to learn how to match the in-camera jpeg engine with your computer based raw converter, and that's only if you use the manufacturer's own converter that includes very precise adjustment intervals. You'll never match the in-camera jpeg perfectly using a third party raw converter unless the third party application has licensed and included in their converter the same algorithms the manufacturer put in the camera's firmware. But that doesn't necessarily mean the in-camera algorithm is either "better" or "worse". It just means it is slightly different. – Michael C Jan 05 '16 at 07:43
  • If they are labeled properly the in-camera jpeg is slightly darker. But, more importantly, has also has higher contrast. The differences in noise reduction and sharpening (both of which also affect contrast) are also readily apparent. Some of the difference in sharpness may also be due to the slight clockwise rotation applied to the raw file. – Michael C Jan 05 '16 at 08:07
2

You should not compare the two formats.

The raw image is the real deal while the jpg is a picture derived(manipulated) from the raw.

Therefore the jpg can never be better than the raw (at least technically).

If the camera made a good jpg, so can the computer ... it's just a question of the processing algoritm the software of your computer/camera have and how much you like the results.

The raw is always superior to the jpg in representing the real picture.

Yet the same rule that let photoshop make a magicial pictures aplies here - the manipulation of a picture can sometimes look better than the original - yet there is always just on e original and it is the raw.

Remember : jpg is the result of computer manipulation over the original picture that tries to create a compressed image as close to the original and as small as possible with the cost of loosing data that the algoritm finds less important

Asaf
  • 561
  • 1
  • 5
  • 11
1

In-camera JPEG could be theoretically better than RAW, in cases where RAW file does not transfer truly raw data from sensor - manufacturers do it to reduce file size ("compressed RAW").

Some cameras, like Sony, only use compressed RAW and this can result in unexpected artifacts in images processed from RAW files, see for example http://www.rawdigger.com/howtouse/sony-craw-arw2-posterization-detection

Of course one may argue that technically this answer is off topic, because compressed RAW is no longer RAW, however people still call it RAW and most of them do not know the consequences.

szulat
  • 5,059
  • 1
  • 20
  • 31
0

Taking as an example the Panasonic DMC-FZ200 I own, several settings are forced to off when recording RAW files even when simultaneously recording JPEG. Among those is i.Dynamic which may scene-dependently choose to dial up ISO sensitivity for better capturing of shadow detail. So the choice to only capture JPEG has consequences for the resulting quality of the JPEG. It's sort of an observer effect situation since given JPEG+RAW from the same shot the JPEG will not contain any information that the RAW file doesn't, while when not even creating a RAW file the JPEG may be better.

0

Yes, in one particular aspect: lens distortion/vignetting/aberration correction. Almost by definition -- and I know for sure on my Nikon -- RAW files are not corrected, while JPEGs are. If your camera has that option, of course.

I've been trying the major free RAW processing programs, and not all support distortion, aberration, and vignette correction -- in general, and for a specific lens in particular.

For example, Rawtherapee reads Adobe lens files, so I get all three with my 18-300mm lens. Marketable uses the open source lens function, which evidently does not have vignette correction for my lens. (Actually, it doesn't actually have my lens at all, so I have to use the 18-200mm lens).

Wayne
  • 491
  • 2
  • 9
  • 1
    On the other hand, Lightroom certainly does support lens distortion/vignetting/aberration correction, so I think saying that this is an advantage of in-camera conversion is a bit of a stretch. – Philip Kendall Jan 01 '15 at 15:13
  • The OP doesn't mention any particular software, and I qualified it with "not all support...". If the OP were using Darktable (which I currently do) and using a Nikon 18-300mm DX lens (as I do), he would not be getting automated vignette correction. The OP was pretty clearly looking for any possible advantage, and I gave him one. It doesn't apply to all software or all cameras (not sure what makes/models do in-camera distortion/vignette correction). – Wayne Jan 01 '15 at 21:17
  • Any math that can be applied in-camera to the raw data during the course of in-camera conversion to jpeg can also be applied exactly the same way to exactly the same data during the course of a post-camera conversion of the raw file to jpeg. The issue you identify with regard to lens correction is not inherent to in-camera vs. out-of-camera conversion of the raw file. Rather, it is an issue due to poorly written raw converters in or out of camera. There are also cameras incapable of in-camera lens correction for which lens correction can be applied using out-of camera converter applications. – Michael C Jan 05 '16 at 07:53
  • @MichaelClark: The question is "Can in-camera..." not "Does in-camera...". Thus my answer is correct. Of course off-camera software can do the same processing -- assuming that: 1) they know the camera as well as the manufacturer, and 2) the saved RAW file has the same amount of information that on-camera software would have access to. But that's a different question. – Wayne Jan 05 '16 at 14:44
-2

You cannot ever have a Jpeg of higher quality than true RAW.

RAW means all information about the picture is there, no cheating, no removing, extrapolating or compressing anything. Each pixel has its color separately defined.

Some companies today do prefer to use size-compression algorithms with RAW pictures, but they should not do that by sacrificing quality. For example, if you have 100 identical Red pixels, then yes, compressed RAW can represent them encoded as something like 100x25500, but this will not reduce quality of it in any way. If quality is reduced, it means the picture is no longer RAW. Even so, due to the way Jpeg format compresses data, theres little chance to have higher quality than any other format used in photography.

Overmind
  • 107
  • 2
-4

In simple terms.....no. As mentioned jpegs are compressed, so they throw away information. Raw has more information. But if the conversion from raw to jpegs is not done right, then your original jpegs may look better. Typically when converting from raw you need to increase contrast and add sharpening.

Namtab
  • 43
  • 4