Jpeg Vs Raw

[dxrocnxj]

well i dont know, i never knew much of a difference between the two formats other than that RAW contained more data, and i dont know know how many of you are subscribed to pop photo, but i read this article in the latest issue, a little more extensive in the magazine but this just show the visual differences between the two.


http://www.popphoto.com/popularphotographyfeatures/4681/digital-toolbox-why-raw-works.html


so put your two sense in or whatever so i can read and learn a bit more.:D

[Prospero]

I am not a subscriber of popphoto, so I haven't read that article.

I found the following article a good read as well: http://www.prime-junta.net/pont/How_to/o_RAW_workflow/_RAW_workflow.html?page=1

It was written as a reaction of the Ken Rockwell article about RAW: http://www.kenrockwell.com/tech/raw.htm

The Rockwell article is really Rockwell at his best... all he says is that you shouldn't use raw because it costs more time and disk space and pros get everything right when they take the shot.

I personally shoot RAW too, I like it for its higher dynamic range, the ability to change settings after taking the shot, and the ability to correct the exposure.
It allows you to recover a lot more detail in the shadows and highlights than is possible with Jpeg.

[erichlund]

Ken Rockwell occasionally says something intelligent. That's important because we all need a shock to our system once in a while. Unfortunately, his opinions on RAW vs JPG are not in that category.

People can shoot jpg all they want. I suppose it's a good way to teach yourself to use the camera correctly, because you learn to fear failure. It's a Pavlov thing. If you are perfect, shoot jpg. No worries. I'm not perfect. I don't always get every camera setting right. I shoot NEF because it's much easier to fix my mistakes. OTOH, just because my Pavlovian response is different from someone else's doesn't make me right. It's just my path through the maze.

[erichlund]

I just read Mr Sulonen's article. He came to many of the same conclusions I did. However, the article is old, and doesn't reflect current technology. I have a laptop and an iPod that I can use to back up my cards, but I get 2GB cards for about $35, so it's easier to get enough cards to cover my needs than deal with the extra step of moving the data to an intermediate hard drive. However, I wouldn't call anyone an idiot for choosing the hard drive route. That's just rude. It's just a choice.

[herc182]

i am not sure how realistic that test in popphoto is. Did the jpeg come straight out of the camera and then they compared it with a RAW photo that you can manipulate to the heavens on photoshop?

If thats the case then its not a fair comparison. You can make a jpeg look nicer if you play with it on the computer....

I always shoot JPEG apart from HDR when i shoot RAW. Apart from that its JPEG all the way. You can even change white balance on lightroom with a jpeg so I dont have to shoot raw for that reason either!

[XaiLo]

I shoot RAW when it counts, if it's just for the heck of it I shoot jpeg and this means a subject that I can always come back and shoot.

[Paradox]

I usually shoot JPEG, because it's quicker to write to the card, takes up less space, and when you know your way around photoshop it doesn't have that much of a disadvantage. However for shots that really count and if you have a few cards with you, it's RAW all the way.

[e_dawg]

JPEG files out of the camera have already been processed to (a) reduce noise, (b) increase sharpness, (c) downconverted from 12 or 14-bits of information per pixel to 8-bits per pixel, and (d) compressed (using lossy compression) to reduce space. All of these steps throw out tonal and subtle colour data and obliterate the original pixel structure. Bottom line, you're not left with much to work with when it comes to post-processing. Once you've thrown away tonal data and pixel structure, it's gone forever. It's like always working with a photocopy instead of the original document.

Every time you save a file as a JPEG, you're throwing away image data because of the lossy compression algorithms. On top of that, JPEG is an 8-bit per pixel data format. That's 256 possible shades of red, green, or blue. Not only is that a quarter of the precision of the original 4096 possible shades for each colour in the 12-bit data from the sensor (or 1/16 of the original 16,384 possible shades for 14-bits), you're not giving your editing program much help by limiting the available precision of its working space. It's like limiting yourself to 1 decimal place when doing hardcore math.

Use TIFF or RAW for your workflow, save the destructive actions like resizing, NR, and sharpening for the end, and leave the JPEG conversion to the last step.

[toriaj]

I've been shooting RAW for maybe 6 months now. I didn't notice much difference when I started, but I figured hey, it's no time at all to write to the card. And I have PLENTY of storage space. So I just didn't move it out of RAW. A few weeks ago I took some "quick" shots in JPEG (the first ones since I started RAW.) I was shocked at how hard they were to work with! I didn't even need to do a lot of editing, but just minor color things to get the pics to have a consistent look. The images started showing artifacts/evidence of the PP much, much quicker than a RAW would have. I finally gave up trying to get them too look right, and just used them the way they were.

So ... try shooting RAW for 6 months, then try to go back, and see what you think! :D

[fionndruinne]

toriaj, are you using color mode II for your jpegs? I always use it, and can still do a fair bit of color editing. However, when I tried mode III, even the smallest amount of fiddling with the blue sky resulted in artifacting. AdobeRGB color profile counts for a lot.

I use raw whenever I know or suspect the scene's dynamic range will be too much for a jpeg. pretty simple, really, from my standpoint.

[Sambru]

I shoot RAW / with JPG Basic - problem solved.:)

[TNB]

I usually shoot RAW (especially since a few mags I know prefer RAW), but I'm shooting for another magazine on Tuesday and they want JPG. For me, I guess it just depends on the situation. ;)

[e_dawg]

toriaj, are you using color mode II for your jpegs? I always use it, and can still do a fair bit of color editing. However, when I tried mode III, even the smallest amount of fiddling with the blue sky resulted in artifacting. AdobeRGB color profile counts for a lot.

fionndruinne,

That seems counterintuitive to theory. When you have a limited amount of discrete values with which to describe the shade of a given colour (in this case, 256, or 8-bits maximum for each colour with JPEGs), the smaller your colour gamut, the better precision you will have in describing said colour gamut, and the better you will preserve said colour information when modifying the image.

Conversely, the larger the colour gamut, the less relative precision you will have because the limited number of discrete values available to describe the colour of a particular pixel will be spread out amongst a much wider range of colours.

An analogy that might help others understand what i am saying is this:

Say the range of all possible colours is geographically equivalent to the size of the US and its territories. Adobe RGB describes the vast majority of those colours in its space, so let's say it's like the US, but without Alaska, Hawaii, Puerto Rico, US Virgin Islands, Guam, and other such protectorates. sRGB, because its colour gamut is not as wide as Adobe RGB, might be equivalent to just the midwest and the east coast*.

If you had only had a Sharpie marker and an 8.5 x 11" piece of paper and had to draw both the entire continental US in as much detail as you possibly could or just the midwest and east coast, the map of just the midwest and east coast would probably be a lot more detailed and precise, relatively speaking, than if you had to draw the entire country.

This is the situation with 8-bit compressed JPEG. There's not a lot of precision there, and when you only have 256 possible shades of each colour within a certain colour space, the colour space with the wider colour gamut will actually be described with less precision than the colour gamut that's smaller.

Furthermore, most monitors can't display colours outside the sRGB colour gamut anyways, and most photo lab printers (Fuji Frontier, Noritsus) can't even render sRGB fully, let alone Adobe RGB (although today's top photo inkjets can render Adobe RGB just fine except for some deep greens).

My opinion is this: if you're using JPEG, use sRGB; if you're using RAW or TIFF, use Adobe RGB.

* - Adobe RGB doesn't actually represent the vast majority of all colours, at least not on the scale I described here, but this is just for argument's sake... sRGB actually represents something like 70% of the colour gamut of Adobe RGB

[fionndruinne]

I see what you're saying. In theory, it makes great sense, but in practice, well, as I said, only a little play with the HSL sliders turned out a harsh gradient right through a blue sky in sRBG, whereas in AdobeRGB I was free to play with the sky a lot in a batch of photos I took on the Oregon coast without a polarizer (thus the sky needed a little adjustment).

Actually, the greater range of AdobeRGB is what allowed me the greater editing flexibility, now that I think about it, for this reason: in the shift in sky color between high blue and the atmospheric color lightening which goes on at the horizon, more grades of color were recorded. Thus in editing that sky, the color channel feature in Lightroom was able to read and alter a large gamut of blue. However, in the sRGB image, there were less colors between dark blue high in the sky, and light blue at the horizon. Thus the software could not maintain the gradual color shift due to having less recorded colors, and a very visible gradient appeared between the range of darker blue and that of lighter.

So your analogy is not perfect, because while the image does describe a "geographical area" in detail, it misses on the transitions between one area and the next, which is something that as either a mapmaker or an image editing program, you need to be able to discern.;)

[e_dawg]

Actually, the greater range of AdobeRGB is what allowed me the greater editing flexibility, now that I think about it, for this reason: in the shift in sky color between high blue and the atmospheric color lightening which goes on at the horizon, more grades of color were recorded. Thus in editing that sky, the color channel feature in Lightroom was able to read and alter a large gamut of blue. However, in the sRGB image, there were less colors between dark blue high in the sky, and light blue at the horizon. Thus the software could not maintain the gradual color shift due to having less recorded colors, and a very visible gradient appeared between the range of darker blue and that of lighter.

Hmm... I think it's actually the other way around if I understand the concepts correctly. Here is an oversimplified example for argument's sake:

Let's say that dark sky is 60% blue and light sky is 20% blue. Those are the "acutal" values of blue of the two ends of the sky.

sRGB has a possible gamut of 0 to 60% blue and Adobe RGB has a possible gamut of 0 to 80% blue.

Working in sRGB colour space

You have to split up the 0-60% blue gamut of sRGB into 256 discrete values. The dark sky at 60% actual vs 60% sRGB gamut is assigned 255 for the value of the blue channel. The light sky at 20% actual vs 60% sRGB gamut is assigned 85 for the value of the blue channel (all I did was 20/60 x 255). In between are 170 discrete values with which to describe the transition between the two shades.

Working in Adobe RGB colour space

You have to split up the 0-80% blue gamut of Adobe RGB into 256 discrete values. The dark sky at 60% actual vs 80% Adobe RGB gamut is assigned 191 for the value of the blue channel. The light sky at 20% actual vs 80% Adobe RGB gamut is assigned 64 for the value of the blue channel. In between are 127 discrete values with which to describe the transition between the two shades.

Discussion

This is how I understand it. Where I think we differ on our understanding is that you seem to be equating the wider colour gamut of Adobe RGB with a greater number of intermediary shades between two actual colours on the spectrum. An analogy would be like: sRGB goes from 1 to 10 in intervals of 1, and Adobe RGB goes from 1 to 10 in intervals of 0.5. Did I infer what you were thinking correctly?

OTOH, i am equating the wider colour gamut of Adobe RGB with the presence of deeper, more saturated colours that do not even exist in the sRGB space. It's like sRGB goes from 1 to 10 in intervals of 1, and Adobe RGB goes from 1 to 20 in intervals of 2. In this case, the universe of possible colours would go from 1 to 30 in intervals of 3. Intervals being defined by the bit depth available, which is the same for all cases. Since the bit depth is the same, when the gamut expands, the intervals become bigger between shades of colour.

Found the link where I read up on this stuff that may help us all, specifically the section on Influence on Bit Depth Distribution:

http://www.cambridgeincolour.com/tutorials/sRGB-AdobeRGB1998.htm

[erichlund]

I usually shoot RAW (especially since a few mags I know prefer RAW), but I'm shooting for another magazine on Tuesday and they want JPG. For me, I guess it just depends on the situation. ;)

You can create jpg from RAW. You cannot create RAW from jpg. If the issue is speed of delivery, then that's understandable, but switching your normal shooting mode just because you have to do a convert doesn't make sense.

[rawpaw18]

Shot this in raw, just been playing with it more and more.

I do not think I could of done such a drastic PP if shot in JPEG.
29969 29970

[TNB]

You can create jpg from RAW. You cannot create RAW from jpg. If the issue is speed of delivery, then that's understandable, but switching your normal shooting mode just because you have to do a convert doesn't make sense.
It has nothing to do with me converting the files since I'm not even involved in the post processing. I also know that they can convert the NEF RAW files to JPG. I also specifically offered RAW files, but they want JPG/LARGE/Fine.

I'll probably just shoot RAW + JPG/LARGE/Fine. However, this would be because I can keep the RAW photos and I still retain the rights to the photos though am unable to submit them to another publication for 180 days.

After that shoot, I have another coming up in which "speed of delivery" is very important and that is because the photos will be downloaded as soon as the CF cards are full.

[erichlund]

I...uh...overstated my case a little bit. :o Didn't really mean to sound that harsh.

[fionndruinne]

This is how I understand it. Where I think we differ on our understanding is that you seem to be equating the wider colour gamut of Adobe RGB with a greater number of intermediary shades between two actual colours on the spectrum. An analogy would be like: sRGB goes from 1 to 10 in intervals of 1, and Adobe RGB goes from 1 to 10 in intervals of 0.5. Did I infer what you were thinking correctly?

OTOH, i am equating the wider colour gamut of Adobe RGB with the presence of deeper, more saturated colours that do not even exist in the sRGB space. It's like sRGB goes from 1 to 10 in intervals of 1, and Adobe RGB goes from 1 to 20 in intervals of 2. In this case, the universe of possible colours would go from 1 to 30 in intervals of 3. Intervals being defined by the bit depth available, which is the same for all cases. Since the bit depth is the same, when the gamut expands, the intervals become bigger between shades of colour.

Yes, you're correct in your interpretation of my standpoint.

I was also going at this with the assumption (correct, actually) that AdobeRGB has a larger volume for recording colors, not an identical amount of space allocated to colors as sRGB does. To quote the article:


Adobe RGB 1998 occupies roughly 40% more volume than sRGB

The writer pointed out how the differences he demonstrates are not as dramatic as he depicted them... my own belief is that in many if not most cases, they're not enough to notice at all, and a large part of that would be AdobeRGB's larger color allocation. Perhaps it's actually the wider range at the ends of specific channels which allows for the more seamless color editing in a blue sky: AdobeRGB's larger cyan channel allows for a less dramatic cutoff between what the program sees as 'cyan' and what it sees as 'blue'. It's often I think in the interplay between colors like this that AdobeRGB becomes important.

[LR Max]

I really think it all depends on what you are using your camera for. When I am shooting for the newspaper, I don't worry about shooting RAW because most likely the image will be B&W and then printed on paper that will obviously degrate the quality. So, whatever. I get it looking pretty good, but that extra little bit will be lost when its printed on paper so why waste my time?

I'll shoot RAW for myself only if its something important or something that I will only have one shot at getting. But I always end up having a good JPG (I shoot RAW+JPG) and I edit that. I think in-camera adjustment is always crucial. RAW is good, but it can't fix blurry images...

[XaiLo]

I want a new colorspace (CWISDspace = Capture What I See (Da...:mad:)space) lol :D

[tcadwall]

Ok... this thread on gamuts might be a bit confusing for some that don't have a sick passion for it... yet.

So here is a strictly mathematical viewpoint.

jpg = 8 bits per color.
2^8 = 256 distinct values per color.
256^3 (3 colors) = 16,777,216 discrete color combinations.

12-bit NEF
2^12=4,096 distinct values per color
4096^3 (3 colors) = 68,719,476,736 discrete color combinations.

Yes, 16.7 Million colors VS 68.7 Billion colors ? ok... the FORMAT allows that.

The power of using a raw image is that colors that YOU CANNOT SEE ON YOUR SCREEN have been captured. But you still can make adjustments that allow you to pull those values into the visible/printable/useable spectrum. I think that the actual storage format is much more capable than the chip that is producing the information!

In other words, the Raw format is way better than jpg regarding the flexibility you have to get the image you want. The format is more capable than your camera.

I am going to have to start a new thread on Windows VISTA .... Hopefully you color pros can help me.

[crum]

a quick couple of questions, if a raw pic is printed does it look poor rather than the jpeg fine equivelent that was also captured? (with a d80) the jpeg choices are a combination of L/M/S and compression fine/med/basic. i understand that L and fine compression will yield the best and S and basic the worst however i was wondering how the other combinations serve and the effective IQ they produce with like-for-like shots? ie is M and fine better or L and med compression?

[K1W1]

You cannot print a RAW picture without first converting it to some other format (typically TIF or Jpeg).
Think of RAW as a negative in film terms.
You use the negative to produce the final image (the print) but the negative is not something that you actually put in the photo album and take around to Grannies.

[Ray Schnoor]

a quick couple of questions, if a raw pic is printed does it look poor rather than the jpeg fine equivelent that was also captured? (with a d80) the jpeg choices are a combination of L/M/S and compression fine/med/basic. i understand that L and fine compression will yield the best and S and basic the worst however i was wondering how the other combinations serve and the effective IQ they produce with like-for-like shots? ie is M and fine better or L and med compression?
S, M & L have to do with the amount of megapixels you are saving the image as. S = 2.5 MP, M = 5.6 MP & L = 10 MP. Unless you are in an emergency situation nearing the end of available memory on your card, always save in L. The same can be said for compression. You can always throw away information by lowering MP and increasing compression after the fact, but you cannot add information you never saved in the 1st place.

Ray.

[K1W1]

You can always throw away information by lowering MP and increasing compression after the fact, but you cannot add information you never saved in the 1st place.

Ray.

Which very neatly takes everybody back to the original RAW vs Jpeg argument. :)

[crum]

thx guys.

K1W1: so even though raw can be viewed in picture project, it can't be accidently printed? i would have thought that 10MP/fine should be handsomely printed at a4? alot of detail seems to be have lost

Ray: wouldn't reduce from L/fine but i was just curious why manufacturers allow this option rather that solely reduce MP or increase compression. i understand that as less MP is captured then it is lost but with increased compression, is detail lost as well? ie it a shot was taken both at L but one with fine and the other with basic compression that the print out would differ in detail and that the largest reproduceable shot would also vary?

[Ray Schnoor]

thx guys.

K1W1: so even though raw can be viewed in picture project, it can't be accidently printed? i would have thought that 10MP/fine should be handsomely printed at a4? alot of detail seems to be have lost

Ray: wouldn't reduce from L/fine but i was just curious why manufacturers allow this option rather that solely reduce MP or increase compression. i understand that as less MP is captured then it is lost but with increased compression, is detail lost as well? ie it a shot was taken both at L but one with fine and the other with basic compression that the print out would differ in detail and that the largest reproduceable shot would also vary?
10MP fine should be able to print photos just fine at A4. There are also several other factors that could be affecting your print. Sharpness, saturation, contrast... are all selectable in the menu if you are saving as a JPG. Set them to your liking. It could also be your printer settings that are the problem.

Yes, as more compression is applied, more detail is lost. Depending on the subject you may or may not be able to detect any difference in printed/viewed output from basic/normal/fine compression. That is just something you will have to decide for yourself.

Ray.

[erichlund]

You cannot print a RAW picture without first converting it to some other format (typically TIF or Jpeg).
Think of RAW as a negative in film terms.
You use the negative to produce the final image (the print) but the negative is not something that you actually put in the photo album and take around to Grannies.

I guess I'm going to have to research how I print my edited RAW files directly from CaptureNX to the printer. Do it all the time.

[crum]

I guess I'm going to have to research how I print my edited RAW files directly from CaptureNX to the printer. Do it all the time.

so it was possible that i printed the raw pic...... should it be drastically different to jpeg if it was taken in auto? if i make adjustments in pictureproject, does it auto save? what is i was just experimenting and want to revert back to the original? there isn't a ok/save or cancel button.

[erichlund]

so it was possible that i printed the raw pic...... should it be drastically different to jpeg if it was taken in auto?
Depends on the software you use. With Nikon CaptureNX, the NEF will look exactly like the jpg if the same settings were used. This is because CaptureNX applies all the settings to the NEF when viewed in CaptureNX (and ViewNX).

if i make adjustments in pictureproject, does it auto save?
I don't use picture project, so I cannot say. Since it's Nikon software, a jpg should appear the same as an NEF (before editing).

what is i was just experimenting and want to revert back to the original?
Again, depends on the software you use. CaptureNX always preserves the original NEF. Jpg is a different story. You have to manually save each generation. Corel Paint Shop Pro auto saves versions if you turn that option on.

there isn't a ok/save or cancel button.

I'm not familiar with picture project, so I cannot help with that. Don't even have it loaded on my computer. My software for editng includes ViewNX (free), CaptureNX ($150?), Corel PSP ($99), and Noise Ninja ($80).

I'm pretty sure ViewNX will do everything Picture Project does, and it's a much better program. It's a free download.

[XaiLo]

picture project is a resource hog that sucks!!! ooops did I say that out loud :D As you can tell no luv lost here that's about one worthless program. what was Nikon thinking :confused:

[tcadwall]

Technically speaking... you cannot print raw.

Some raw editors will allow you to print, but they are converting for the print. That might be confusing, but true.

We keep talking about jpg and Raw... Which of course is the title of the thread.... However, there IS another option for prints. Tiff. I have successfully converted raw to 16bit tiffs that have gone to the lab for prints. I normally only do this for b/w prints since you using a very limited color pallette with b/w jpgs.

Of course, this can only be done within reason - depending on the processing power at the lab!!!!

I took in a b/w panoramic shot in 16bit TIFF.... I can't remember exactly but the rez was somewhere around 5100x1500 and the file size was somewhere in the neighborhood of 45MB!!!! I crashed their processing software! About 10 prints came out double exposed, squiggly, and almost indescernable.:eek::D

[crum]

picture project is a resource hog that sucks!!! ooops did I say that out loud :D As you can tell no luv lost here that's about one worthless program. what was Nikon thinking :confused:

not only me that thinks pictureproject could be better..........is capture nx the only alternative that is free? not sure i wan to spend more money on pro software at the moment.........any reason why nikon dun bundle capture nx as well?

oops! i meant view nx as mentioned earlier.

[tcadwall]

NX is free?

[erichlund]

ViewNX is free. You just have to download it from Nikon or buy a D3/D300.

CaptureNX is NOT free. However, I believe it is being bundled with D3s and D300s. Not sure where. It was being bundled with D200s in Britain, though I don't know if that's still on.

[erichlund]

Technically speaking... you cannot print raw.

Some raw editors will allow you to print, but they are converting for the print. That might be confusing, but true.

We keep talking about jpg and Raw... Which of course is the title of the thread.... However, there IS another option for prints. Tiff. I have successfully converted raw to 16bit tiffs that have gone to the lab for prints. I normally only do this for b/w prints since you using a very limited color pallette with b/w jpgs.

Of course, this can only be done within reason - depending on the processing power at the lab!!!!

I took in a b/w panoramic shot in 16bit TIFF.... I can't remember exactly but the rez was somewhere around 5100x1500 and the file size was somewhere in the neighborhood of 45MB!!!! I crashed their processing software! About 10 prints came out double exposed, squiggly, and almost indescernable.:eek::D


If we are getting technical, I don't think the printers print jpg or tiff either (I could be wrong). I believe the print drivers convert the images to a bitmap, which they can print. Epson photo printers come with a print driver called RAW Print, but I don't know how up to date it is. I don't recall actually trying to use it.

Another application, QImage, is directly designed for printing photos. It's main claim to fame is it's ability to Upres without jaggies and print large. It will upsize and print many RAW formats in its Studio edition ($90). BTW, if you really want to print large, this is one of those programs to consider.