Discussion
Is RAW really 'that much better' than JPEG? 
RAW is the same image at the same resolution, but without jpg compression.
Im not challenging the concept of raw being better, but superfine .jpg's are so "uncompressed" as they are, flawless enough to the naked eye, and come out with 0.5% dataloss but 400% reduction in file size.
What do you shoot in, under which circumstances, and why.
>> Edited by UKBob on Tuesday 7th February 17:09
RAW is the same image at the same resolution, but without jpg compression.
Im not challenging the concept of raw being better, but superfine .jpg's are so "uncompressed" as they are, flawless enough to the naked eye, and come out with 0.5% dataloss but 400% reduction in file size.
What do you shoot in, under which circumstances, and why.
>> Edited by UKBob on Tuesday 7th February 17:09
Think of RAW as your negative, and JPG as your print.
If you shoot a roll of film and take it into Snappy Snaps to get developed and printed, they'll give you back a set of prints and the negatives.
The prints would have been processed in a particular way and at a particular size from the original negatives, so if you wanted to alter the colour or exposure slightly, or just print them bigger, you would get better results by doing this from the original negatives rather than working with the print.
When you set a camera to output jpg files, it is in effect "developing" your prints with the settings that you configure (colour saturation, sharpness and size for example), but the RAW files are the digital version of the negative and can be used to create new "prints" outside the camera.
If you shoot a roll of film and take it into Snappy Snaps to get developed and printed, they'll give you back a set of prints and the negatives.
The prints would have been processed in a particular way and at a particular size from the original negatives, so if you wanted to alter the colour or exposure slightly, or just print them bigger, you would get better results by doing this from the original negatives rather than working with the print.
When you set a camera to output jpg files, it is in effect "developing" your prints with the settings that you configure (colour saturation, sharpness and size for example), but the RAW files are the digital version of the negative and can be used to create new "prints" outside the camera.
My RAW files are 24MB, but thats all I shoot in. I think Ed's description was spot on, and that's the reason I only use jpegs for interweb use.
Martin.
Edited to add:- the image bank I submit to want 48MB images minimum, another reason for not compressing .
]
>> Edited by V6GTO on Tuesday 7th February 18:57
Martin.
Edited to add:- the image bank I submit to want 48MB images minimum, another reason for not compressing .
]
>> Edited by V6GTO on Tuesday 7th February 18:57
Imperfect 'in camera' jpeg processing: Priceless
Cost of 5MB disk space for compressed RAW file: Not a lot
Simple equation for me.
Add to that the fact that you need RAW to generate good quality TIFF's or any other format you like - as required by most publishing houses - and its a no-brainer for me.
Cost of 5MB disk space for compressed RAW file: Not a lot
Simple equation for me.
Add to that the fact that you need RAW to generate good quality TIFF's or any other format you like - as required by most publishing houses - and its a no-brainer for me.
Another good explanation here:
www.luminous-landscape.com/tutorials/understanding-series/u-raw-files.shtml
www.luminous-landscape.com/tutorials/understanding-series/u-raw-files.shtml
RAW gives you more options, but on my 350d etc the in camera processing isnt to bad and JPG quality has been seen (steves digi's etc) to have almost identical quality to the raw.
If your into post processing/workflow much then shoot raw, if you just want out of the camera images find some settings your happy with (white balance, sharpness , colour etc) and go for it.
Not sure why someone would only want a 48mb image even if the raw was smaller unless your submitting a ps file with layers etc...
If your into post processing/workflow much then shoot raw, if you just want out of the camera images find some settings your happy with (white balance, sharpness , colour etc) and go for it.
Not sure why someone would only want a 48mb image even if the raw was smaller unless your submitting a ps file with layers etc...
GetCarter said:
Once you shoot in jpg - you are stuck with it.
I know what you mean but you can still spin it to hell and back in Photoshop or similar, adjust levels, contast, colour balance, sharpen, transform, apply filters etc etc. For most applications and most people that's probably enough.
I use JPG for 'normal' stuff but RAW for demanding or important situations, in particular high contrast subjects and awkward white balance. I use Adobe Camera Raw (ACR), my only niggle being that there's no 'Levels' control and I can't get the image full screen.
>> Edited by simpo two on Tuesday 7th February 20:15
Take two identical exposures, one in jpg (any mode you like) and one in RAW.
Try to get am image with some nice slowly changing colour/brightness intensity changes, like a nice blue sky - darkening off to the top (or use a polariser) . . . or some nice foggy pictures . . .
Put both into photochop and convert both to TIFF.
Now have a play with levels on both images . . . notice how the jpg image develops lots of horrible banding in the slowly changing graduations whereas the RAW doesn't.
That's your 'insignificant' data loss . . .
A jpg images is compressed to look correct without any changes - as soon as you make changes the compression is wrong and may have thrown away data it may otherwise have kept.
I took a lot of jpg images when I got my D70, I deeply regret not using RAW now . . .
Still - it depends what's important to you . . .
I use jpg when I do not intend to do any/much post processing . . .
Fd
Try to get am image with some nice slowly changing colour/brightness intensity changes, like a nice blue sky - darkening off to the top (or use a polariser) . . . or some nice foggy pictures . . .
Put both into photochop and convert both to TIFF.
Now have a play with levels on both images . . . notice how the jpg image develops lots of horrible banding in the slowly changing graduations whereas the RAW doesn't.
That's your 'insignificant' data loss . . .
A jpg images is compressed to look correct without any changes - as soon as you make changes the compression is wrong and may have thrown away data it may otherwise have kept.
I took a lot of jpg images when I got my D70, I deeply regret not using RAW now . . .
Still - it depends what's important to you . . .
I use jpg when I do not intend to do any/much post processing . . .
Fd
If you plan to only shoot for web use JPEGS are fine, Raw files allow much more flexibility. I understand that the JPEG compression does change the image every time you open and save it. If you need really large images for adverts etc then the only starting point is a nice large raw file!
fergusd said:
I took a lot of jpg images when I got my D70, I deeply regret not using RAW now
That's it right there. As I get better at post processing, and as I've realised more the flexibility that RAW offers, I often go back over old images seeing if I can improve them.
Except the ones I shot in jpeg which I also deeply regret.
Something else that doesn't appear to have been mentioned is that IIRC JPEG is limited to eight bits per channel (Source Adobe Camera Raw Book by someone who's name escapes me at the moment). If this is OK, when why are we all splashing out on cameras that do at least twelve bits?
My personal experience is a Fine, High Quality jpeg from my 20D. This was taken straight out of the Camera - manipulated only slightly to turn it to a sort of dark sepia (to match our bed), and then printed (fantaistically btw) by Fluid Images onto a 1.1m canvas.
The definition and sharpness of most of the print is fantastic, but the edges of the clouds show a degree of jpeg-ish-ness. Gutted as I can neveer retake the shot.
Camera now switched over permanenetly to RAW.
Rich
My personal experience is a Fine, High Quality jpeg from my 20D. This was taken straight out of the Camera - manipulated only slightly to turn it to a sort of dark sepia (to match our bed), and then printed (fantaistically btw) by Fluid Images onto a 1.1m canvas.
The definition and sharpness of most of the print is fantastic, but the edges of the clouds show a degree of jpeg-ish-ness. Gutted as I can neveer retake the shot.
Camera now switched over permanenetly to RAW.
Rich
simpo two said:
Also don't forget the RAW allows you to adjust the exposure up to 4 stops in either direction (depending on the software you use).
What about noise as you lighten the shadows?
Mind you, if I'm four stops out I'd rather go and weash cars for a living!
The noise is nowhere near as bad as if you used a JPG.
True, 4 stops is very extreme, I have never shoved anything more than about 2 stops and then Noiseware comes in handy sometimes...
So does RAW save more light intensity information then?
I'm kind of confused what people here are trying to say as a friend is starting in photography and questioned me about this.
JPEG will save 256(2ˆ8bit) intensities for each channel, red, green and blue, (so 24bit) and then compress them using defined quality compressor.
What does RAW save? I read 12bit per channel earlier, so that would be 4096(2ˆ12bit) per channel of red green and blue.
Is that where the benefit comes in?
How is this information stored though? Does the camera simply collect a larger range of intensity of colours from the real world and then toss alot of it away and generate a JPG, or does it save this excess information in the RAW?
OR, does the camera simply catch a snapshot of light information, and simply store it either as JPG or RAW, with the RAW storing the same dynamic range, but with more instances of selectable information within that range which the JPG format rejects?
Cheers
Dave
I'm kind of confused what people here are trying to say as a friend is starting in photography and questioned me about this.
JPEG will save 256(2ˆ8bit) intensities for each channel, red, green and blue, (so 24bit) and then compress them using defined quality compressor.
What does RAW save? I read 12bit per channel earlier, so that would be 4096(2ˆ12bit) per channel of red green and blue.
Is that where the benefit comes in?
How is this information stored though? Does the camera simply collect a larger range of intensity of colours from the real world and then toss alot of it away and generate a JPG, or does it save this excess information in the RAW?
OR, does the camera simply catch a snapshot of light information, and simply store it either as JPG or RAW, with the RAW storing the same dynamic range, but with more instances of selectable information within that range which the JPG format rejects?
Cheers
Dave
I'll have a go at this, although someone better able to get the point across will be along shortly no doubt.
There are two separate issues. Firstly file format:-
RAW is just that - it is a file containing the raw data that the camera collected. I imagine (although am not sure) that it will compress the information using a lossless system (such as zip on a pc), but essentially is an array of data giving the light intensities at each pixel and each channel. As the formats are camera specific, they will store all the info collected by the camera.
This information is essentially useless in the real world, as the sensor sees light linearly, whereas we don't - so gamma correction is required in the PC to turn the data back into an image that we would recognise.
JPEG is a lossey compression technique - developed by some cleaver people that understand how we take information out of a picture. It looks (I believe) at changes of tone between pixels etc and sees how much of the information that it can throw away without us noticing. This gets the massive compression rates - we may only halve the size of a word file when we zip it, but a JPEG may only be 10% of the size of the associated tiff.
The bit depth issue is slightly different - JPEG only caters for eight bits per channel, therefore it cannot store any more information than this. All images are RAW when they come off the sensor. The camera then applies its own gamma correction (to turn the raw data into a digital picture), then applies JPEG compression. It doesn't matter what the bit depth of the source is, the JPEG file format can only contain eight bits per channel - so essentially the data is thrown away (although I'm sure the process is slightly more complex)!!!!
From reading the responses on this and other threads, I think that the bit depth issue is a bit of a red herring (although bigger is usually better). The main complaints come in the fact that the JPEG compression process always leaves certain witnesses that you can see. Also, every time you save a jpeg, the compression process happens again, so you lose more data. Having less data to begin with also limits how much manipulation you can do without the witnesses becomming greater.
I've only been caught out once, out of a fair few photos taken. But, as I mentioned above, I can never go back and get the photo back - carrying a bit more memory seems to be a small price to pay to know that you can go back to the original data that the camera captured at any time in the future
There are two separate issues. Firstly file format:-
RAW is just that - it is a file containing the raw data that the camera collected. I imagine (although am not sure) that it will compress the information using a lossless system (such as zip on a pc), but essentially is an array of data giving the light intensities at each pixel and each channel. As the formats are camera specific, they will store all the info collected by the camera.
This information is essentially useless in the real world, as the sensor sees light linearly, whereas we don't - so gamma correction is required in the PC to turn the data back into an image that we would recognise.
JPEG is a lossey compression technique - developed by some cleaver people that understand how we take information out of a picture. It looks (I believe) at changes of tone between pixels etc and sees how much of the information that it can throw away without us noticing. This gets the massive compression rates - we may only halve the size of a word file when we zip it, but a JPEG may only be 10% of the size of the associated tiff.
The bit depth issue is slightly different - JPEG only caters for eight bits per channel, therefore it cannot store any more information than this. All images are RAW when they come off the sensor. The camera then applies its own gamma correction (to turn the raw data into a digital picture), then applies JPEG compression. It doesn't matter what the bit depth of the source is, the JPEG file format can only contain eight bits per channel - so essentially the data is thrown away (although I'm sure the process is slightly more complex)!!!!
From reading the responses on this and other threads, I think that the bit depth issue is a bit of a red herring (although bigger is usually better). The main complaints come in the fact that the JPEG compression process always leaves certain witnesses that you can see. Also, every time you save a jpeg, the compression process happens again, so you lose more data. Having less data to begin with also limits how much manipulation you can do without the witnesses becomming greater.
I've only been caught out once, out of a fair few photos taken. But, as I mentioned above, I can never go back and get the photo back - carrying a bit more memory seems to be a small price to pay to know that you can go back to the original data that the camera captured at any time in the future
Gassing Station | Photography & Video | Top of Page | What's New | My Stuff