JPEG - And Data Loss

[graphilly]

Well to me it looks pretty obvious without even enlarging the attached image, but what do I know

I went for right. It has more vibrant colours and it looks sharper.

 

[simonblue]

Left for me

 

[weybourne]

I have gone to the largest picture in Flickr and zoomed in as much as my browser will go - and I cannot tell the difference - sometimes I think I can see a difference in the artefacts and then I think I can't or that I'm imagining it. Most of the time I see exactly the same artefacts on both images.

Artefacts aside, the image on the right looks a tad sharper and more vivid when I look at the crowds.

 

[dinners]

This exercise is about degree of degradation of jpeg images after successive saves and is nothing to do with the merits (or not) of shooting jpeg over raw.

I agree

I shoot jpeg most of the time but I understand it's limitations if you don't get it 'just about right'.

Nail it in jpeg and you don't have to worry.

Fail in jpeg and the Raw comes into it's own.

In terms of the multi save point though - a good read.

 

[ZoneV]

I'm surprised by the lack of difference between the two images. I would have expected to see more banding in the subtle graduations of colour in the smoke/bright light (more like the obvious banding in the water of Pookeyhead's example).

Regarding workflow, if I was shooting JPEG the first thing I'd do as soon as the files were uploaded to a computer (after binning obvious duffers) is save them as a non-lossy format such as TIFF (with compression set to "none"). That way the original JPEG is safe, and the non-lossy TIFF file can be edited and re-saved as many times as necessary without any loss.

I almost always shoot in RAW, and when I switched to Lightroom I found its workflow and file/version management made things much easier than the traditional method of "saving as" and creating a new duplicate file each time. Because Lightroom in "non-destructive" (i.e. it never changes the image file), I always work from the original RAW, only exporting as a JPEG when ready to output the final image.

 

[EOS_JD]

I've done the same with a portrait image i have and it's clear which is which as the pixels have started breaking up on the skin. Saving a high res jpg is unlikely to do any major damage to your files though.

 

[Pookeyhead]

I'm sorry, but there's something goig wrong with the OP's test.

I used the same image, as a starting point..

I then loaded it, used file/save as and saved it as a JPEG, max quality, then closed the image. I then repeated that, using file/save as JPEG max quality, closed it etc... and I did that only 10 times. The results are below.

So. This is exactly what I found with my Orkney image further up the thread.

I'm not doing anything weird... I'm opening, going to file, save as, and selecting JPEG Max quality, then closing the image.

Look at the sky in the second one! I have no idea what Matt's doing, but it's not being re-compressed each time he saves like mine are.

 

[Jan K.]

It maybe depend of which editor you use?

 

[Pookeyhead]

It maybe depend of which editor you use?

Maybe it does. Maybe some keep the image cached... (shrug). I'm using Photoshop CS6 though.... the industry standard image editor.. as most probably are.

Seriously.. I'm not trying to be an arse... When I re save JPEGs, they're noticeably degraded after 3 saves, and utterly ruined by the time I get to 10 re-saves.

 

[DuncanDisorderly]

David - did you use Matt's original files or the smaller image posted in this thread?
I'm asking because the original has shadow noise which can be seen at 100%.
This noise helps hide the banding. The smaller image presented in this thread has smooth tones and hence will band far more easily.

 

[killwilly]

I voted for the one on the right as the single edit, simply because the colours are much stronger on my screen. However, I think the left image is very acceptable.

 

[MWHCVT]

That's incredible, for clarity I processed the RAW in CS6 ACR, but when it came to the opening save, then open the new file save etc I used Freeware PhotoScape, I see no reason why I'd get such different results, and what I would suggest are more concerning results from the software that has cost me several hundred pounds one would have thought that the paid for software would do the best job

 

[markyboy1]

This exercise is about degree of degradation of jpeg images after successive saves and is nothing to do with the merits (or not) of shooting jpeg over raw. RAW files contain more information than jpegs and so give you more data to play with before saving as, for the sake of this argument, a 'first generation' jpeg - that extra information is 'discarded' by the camera when you shoot jpegs straight out of the camera. If you are happy with the jpegs sooc, then there is no point shooting raw. If you want to turn you hand at PP on the computer, then there are distinct advantages in working with the raw data to 'craft' your jpeg.

I got the point of the exercise Adrian and how the camera works with regard to in body conversion. My point was simply that although no editing was done during the 50 saves I cant tell enough difference to promote the sole use of Raw.
That said I will still continue to shoot my shots in Raw because I like been able to fiddle my images and it helps me to understand the PP a little better.

Mark

 

[EOS_JD]

I'll post my results with a 5DII file later if I get a chance. There's a clear difference.

The flaw above is that you're showing the whole image which is reduced both is size.

Show us 100% crops of each...... That will show the difference I'm sure.

 

[MWHCVT]

I'll post my results with a 5DII file later if I get a chance. There's a clear difference.

The flaw above is that you're showing the whole image which is reduced both is size.

Show us 100% crops of each...... That will show the difference I'm sure.

I don't believe in 100% crops, you don't naturally look at a photo like that you view them as a wider image

 

[Slaphead]

I'm going with left - there are differences but you need to look closely at the full res version.

What makes me think it's the left is that the streaks from the fireworks are sharper in the right version. In the left version they have a slightly blue halo.

Also in one of the small castle windows I can see a bit of CA in the right image, but in the left the CA is all but gone.

There are other differences as well which are especially visible in the areas where there are lights.

Of course on printed media, or viewing at a normal size the differences will be negligible.

I would have thought the differences would have been far more apparent after 50 saves.

 

[DazJW]

Indeed it would One of the things for me the colours are a little more punchy in the original to my eyes

That was the third factor for me, in one area in particular

I don't believe in 100% crops, you don't naturally look at a photo like that you view them as a wider image

I think it's more about safety margins. Obviously some people go crazy over the top about it but if you know something looks fine at 100% it'll look increasingly better at everything less than that and you can get on with doing more interesting things without worrying about it. It frees you from having to give any further consideration to it.

 

[EOS_JD]

I don't believe in 100% crops, you don't naturally look at a photo like that you view them as a wider image

Correct BUT if you are trying to show the effect of jpg compresion you HAVE to show the image at 100% to see it.....

You are showing two images above that when printed would be about 2" high! you can't see ANY detail at that size.

So it's really down to how large you are going to print any image. Also when you downsize the image you're probably throwing away most of the rubbish pixels anyway!

I think you HAVE to show 100% crops to show the effect. Anything else is false.

 

[MWHCVT]

Correct BUT if you are trying to show the effect of jpg compresion you HAVE to show the image at 100% to see it.....

You are showing two images above that when printed would be about 2" high! you can't see ANY detail at that size.

So it's really down to how large you are going to print any image. Also when you downsize the image you're probably throwing away most of the rubbish pixels anyway!

I think you HAVE to show 100% crops to show the effect. Anything else is false.

Yes and you can click the link and look at them at the full size original I can take a 100% crop of some areas where there is little to no difference and some where there are lots of difference

 

[EOS_JD]

I agree and that is down to the way compression works. And some images will be better than others depending on what is contained.

Where is the full size link? I only see a link that is 1024px x 640px which is a very low resolution when it contains two images.

Edit - I clicked the link and see that the size is now 2048 x 1280 - So still a very small image that is 1280 x 1024 - Would produce a print very slightly larger than 4" x 3" of each image!

Surely the camera you used is higher resolution than that? Again you've thrown away much of the data - Although I agree it's hard to tell the difference.

 

[MWHCVT]

I agree and that is down to the way compression works. And some images will be better than others depending on what is contained.

Where is the full size link? I only see a link that is 1024px x 640px which is a very low resolution when it contains two images.

In my OP copied here for ease of use

And if you want to look closely then CLICK HERE to go the original full size file

 

[ZoneV]

I'm sorry, but there's something goig wrong with the OP's test.

I used the same image, as a starting point..

I then loaded it, used file/save as and saved it as a JPEG, max quality, then closed the image. I then repeated that, using file/save as JPEG max quality, closed it etc... and I did that only 10 times. The results are below.

So. This is exactly what I found with my Orkney image further up the thread.

I'm not doing anything weird... I'm opening, going to file, save as, and selecting JPEG Max quality, then closing the image.

Look at the sky in the second one! I have no idea what Matt's doing, but it's not being re-compressed each time he saves like mine are.

That's more like it. Looks like "save as" does the damage, hence why Matt's don't look bad.

 

[Suvv]

That's more like it. Looks like "save as" does the damage, hence why Matt's don't look bad.

See post #5
Some editors are better than others when you do this.

 

[ZoneV]

So it seems. This explains it:

"... some image editors do recompress JPEGs when the Save As command is used . To avoid this you should duplicate and rename JPEGs in a file manager rather than using "Save As JPEG" in an editing program."

http://graphicssoft.about.com/od/formatsjpeg/a/jpegmythsfacts.htm

 

[MWHCVT]

That's more like it. Looks like "save as" does the damage, hence why Matt's don't look bad.

It's funny as I used "save as"

 

[ZoneV]

It's funny as I used "save as"

You did? I see, the confusion arises from your earlier post which gave the impression you just used "save":

" it was an open save, re open and save 50 times"

If you were indeed using "save as" then the image editor you use must be one that doesn't recompress during a "save as" as mentioned in the piece I quoted, hence the lack of degradation in your example when compared to Pookeyhead's (which is more typical of what you'd expect a heavily re-saved JPEG to look like).

 

[MWHCVT]

You did? I see, the confusion arises from your earlier post which gave the impression you just used "save":

" it was an open save, re open and save 50 times"

If you were indeed using "save as" then the image editor you use must be one that doesn't recompress during a "save as" as mentioned in the piece I quoted, hence the lack of degradation in your example when compared to Pookeyhead's (which is more typical of what you'd expect a heavily re-saved JPEG to look like).

Well it has to be save as I didn't think to type has save isn't even an option if all you've done is open the file I would have though most people who can turn a computer on would get that

 

[ZoneV]

Well it has to be save as I didn't think to type has save isn't even an option if all you've done is open the file I would have though most people who can turn a computer on would get that

It always helps to explain things properly, Matt, especially when the method of saving is the crux of what you're demonstrating.

Anyway, I'll leave you to play around with your "tests", I think most people know the disadvantages of working with lossy formats.

 

[EOS_JD]

I think most people know the disadvantages of working with lossy formats.

Steer clear of jpg to start with (use for printing is fine). Saving a high res image a couple of times even at quality 10 is unlikely to damage the file much but repeated closing and saving will.

 

[modchild]

Most of my photos are for personal pleasure and not taken with a sell on value and for these I shoot jpegs, at an airshow over a weekend I can often shoot 3000 + photos and I really can't be bothered to do PP on all of them, specially as most of the shots I display are on a digital photo frame.

For any commissioned photos I shoot in raw + jpeg or just raw so I can get the very best out of them for the customer, although the last 2 times I shot in raw + jpeg the customer preferred the original jpeg to my PP'd raw shots. I can honestly say I've never edited and saved any of my photos more than 3 times and usually they get only 1 edit and then saved.

All of this "you should only shoot raw" or "raw is best" just annoys me. We end up with so many threads about it that I just don't read through them, this thread being a different angle on a familiar theme was one I read all the way through. However many times people 'tell' other people what to do isn't going to change many minds. I don't see either way being better overall, but both are better in certain circumstances.

And for the record, I think the right side image is the 'saved once' one. To me it just looks brighter to me. And a wonderful landscape photo in the second comparison David, that's a cracker.

 

[EOS_JD]

Stuart
Sorry but it's a fact that RAW holds more data than a jpg. Shoot what you like yes but be aware that a RAW image is a better file - if you have the time or need to use it.

I'm not saying anyone HAS to use it but really space is so cheap it seems a waste not to. I've shot at airshows and RAW has saved a few cracking shots from the bin - where they would have been had I shot jpg!

The fact your client preferred the original JPG to the processed raw couldsay more about your processing skills? Not trying to be cheeky - just wondering as that's not much of an argument not to shoot in RAW.

There's always two sides though and if you are assured of getting exposure bang on with every shot, jpg could work out fine.

 

[phil8139]

Thought this wasn't going down the line of what to shoot in?

For the record I have not shot a raw file in 2 years, never had an issue in choosing the right image to edit and not using expensive image editors probably helps too!

Would be good to see if the image loss is different on different editors :shrug:

 

[EOS_JD]

Here's 2 images 100% crops. "Save As" 12 quality JPG, closed opened and repeated 50 times. I created a small action to do that lol

This to me though is a bit of a waste of time as no-one is going to save over and over like this and JPG images will hold up to the few saves that some may do.

Should be easy to see the loss of detail.

 

[phil8139]

So...... in my humble opinion, the question of loss is a valid one, but the amount of loss in average conditions is hardly worth the stampede to shoot raw or raw & jpeg. Use the system that suits yourself.

It seems the same argument in buying branded beans or value beans, they will do the job in filling your hunger spot and your farts will probably still smell the same

 

[AHILL]

A very Interesting thread Matt.

I hope I'm not throwing a spanner in the works but would degredation through compression be the same regardless of the editor being used?
ie: Coreldraw/ photopaint, Photoshop or freeware programs.

 

[MWHCVT]

Here's 2 images 100% crops. "Save As" 12 quality JPG, closed opened and repeated 50 times. I created a small action to do that lol

This to me though is a bit of a waste of time as no-one is going to save over and over like this and JPG images will hold up to the few saves that some may do.

Should be easy to see the loss of detail.
https://dl.dropbox.com/u/11462887/IMG_14191.jpg
https://dl.dropbox.com/u/11462887/IMG_1419.jpg

Yeah that's very clear

 

[realspeed]

removed

 

[modchild]

!

The fact your client preferred the original JPG to the processed raw couldsay more about your processing skills? Not trying to be cheeky - just wondering as that's not much of an argument not to shoot in RAW.

I'm the first to acknowledge I'm far from the best when using PP, but mainly using auto fix in PSE9 and auto levels in Canon DPP they still preferred the jpegs. Having said that, they were the only one out of all the people who saw the photos that preferred the jpegs. Just their taste I suppose. I have done raw shots at many events to see if I can improve the look by much, but I really like the jpeg IQ out of my 5D3, and by the time I've finished there doesn't look much in it.

These raw vs jpeg arguements will go on for as long as camera manufacturers make the technology available, but really the are about as pointless as the Manual only vs AV/TV/P/ debates. Most people wont change what they shoot because someone else is telling them to.

Jim, what was the image like after only 2 or 3 changes and saves. I've got over 40K jpeg photos on my hard drive and I've never needed to alter any of them more than 3 times at the most, normally only once. By that amount the quality loss will be negligable anyway.

By the way, thanks Matt for taking the time for the comparisson in the first place.

 

[Pookeyhead]

David - did you use Matt's original files or the smaller image posted in this thread?
I'm asking because the original has shadow noise which can be seen at 100%.
This noise helps hide the banding. The smaller image presented in this thread has smooth tones and hence will band far more easily.

I used the ones he posted on here. I've found the same with my own images though. By the time I reach 10 re-saves, it's ruined. I've no idea how Matt managed 50 saves with so little degradation, but I can not replicate it. It is also not consistent with how JPEG compression works.

 

[specialman]

couldn't tell looking at the samples.

how much difference is there at a more humble number of saves, say three times?