jpeg vs "lossless" formats

Humphrey Nash said:

I'm not sure I understand your point. My experiment was to save an original large JPEG file with the name "Saved 1" at 75% quality. This file was roughly 70% smaller than the original file (with no visible IQ difference). I closed the JPEG and opened up the "saved 1" and then saved that file as "saved 2" using 75% quality. I repeated this process until I had a fourth generation "saved 4" file.

Click to expand...

In your previous description, you hadn’t indicated that you closed the JPEG and reopened it. You simply said “repeatedly saved”.

Humphrey Nash said:

All JPEGs after the first compression were the same size, i.e., no further compression.

Click to expand...

Just because they're the same size doesn't mean that detail wasn't lost. This is evident by the fact that you can simply open a JPEG, save it at a higher quality level, and get a larger file. Now, did selecting a higher quality level increase detail? No, that's impossible. So there must be more going on with the JPEG compression scheme than meets the eye.

Humphrey Nash said:

I even examined the individual pixels at 1000% (interpolation turned off) and there were NO differences after the first compression.

Click to expand...

Did you perform a difference blend? That will show you exactly where the differences are, and the RGB values will tell you by how much.

Humphrey Nash said:

BTW, using two of my programs (Compupic and Paint Shop Pro) saving a JPEG at the highest quality (100% or no compression) INCREASES the file size substantially yet when I pixel peep at 1000% there is NO difference.

Click to expand...

It’s not a puzzle at all if you understand how JPEG compression works. Saving at 100% will still cause the image to degrade. As I said before, lost detail is an unavoidable byproduct of the JPEG compression process.

You seem to have a preconceived notion as to how JPEG compression works that's not correct. I suggest you read up on it if you’re interested.

http://en.wikipedia.org/wiki/JPEG

.

Humphrey, can you give an opinion re compupic compared to Faststone please, if of course you have used the latter.
I find Faststone much quicker for browsing.

Carl

I have no experience with Faststone. I use Compupic because it is flexible yet simple, a good match for me. In addition to being a viewer for various formats it allows me to do most (95%) of my PP easily. I consider it ideal for pixel peeping if that's of any interest to you. For example the following (10X uninterpolated) is fairly easy to produce but I haven't a clue how to do it with other programs.



Compupic has a free trial. Note that this program has some advantages over, for example, Google's Picassa but is from a small company so it has limited functionality.

This HAS to be the most under-rated image processing software ever (and it's free):

http://www.getpaint.net/

I use Capture One, LightZone and Olympus Studio .. and NONE of these programs are as convenient as Paint.net. It's fantastic!

Cheers,

http://www.pbase.com/plusiotis/insect_freaks_and_aberrations

but the hole is so small that you can't see any petrol drip onto the pavement, does that mean you don't have a hole?

JPEG is by definition lossy however, the loss might not be significant enough for a human to judge visually depending on the scene (uniform blue sky vs more busy background), technique, assessment method and so on.

But JPEG leaks "by definition"

So, one could say "JPEG is good enough for me" but one cannot say "JPEG does not leak"

Graystar said:

You seem to have a preconceived notion as to how JPEG compression works that's not correct. I suggest you read up on it if you’re interested.

http://en.wikipedia.org/wiki/JPEG

.

Click to expand...

--



Ananda
http://anandasim.blogspot.com/

Good point Timo. I have seen blocky blue skies and my pictures don't illustrate that JPEG failing.

Having said that a thought occurs to me. I haven't seen those blocky or serrated blue skies for some long time (years). I wonder if new in-camera processing has cleared up that issue. Or it may be that today's cameras with their higher pixel counts don't show that JPEG blockyness. I have never seen it in any of my pictures and I shoot JPEGs almost exclusively.

This raises some interesting questions. I assume that blockyness, when and if it occurs, must be JPEG failing and not present if you shoot RAW? But when you display a RAW image it is converted to JPEG so the PP conversion must avoid blockyness. This implies that the in-camera RAW conversion to JPEG, not JPEG per se must be the source of blockyness. Maybe this is the reason newer cameras don't seem to have this problem, i.e., their in-camera RAW converters have caught up with PP RAW converters.

Humphrey Nash said:

Good point Timo. I have seen blocky blue skies and my pictures don't illustrate that JPEG failing.

Having said that a thought occurs to me. I haven't seen those blocky or serrated blue skies for some long time (years). I wonder if new in-camera processing has cleared up that issue. Or it may be that today's cameras with their higher pixel counts don't show that JPEG blockyness. I have never seen it in any of my pictures and I shoot JPEGs almost exclusively.

This raises some interesting questions. I assume that blockyness, when and if it occurs, must be JPEG failing and not present if you shoot RAW? But when you display a RAW image it is converted to JPEG so the PP conversion must avoid blockyness. This implies that the in-camera RAW conversion to JPEG, not JPEG per se must be the source of blockyness. Maybe this is the reason newer cameras don't seem to have this problem, i.e., their in-camera RAW converters have caught up with PP RAW converters.

Click to expand...

I've always assumed, rightly or wrongly, that it was a question of the degree of compression. It is certainly exacerbated by over-sharpening, that accentuates small tonal differences between adjoining areas, in this case jpg 'blocks'. I think one reason it's not so obvious in straight raw conversions might be that sharpening has not been applied to the same extent.
--
tim
http://www.pbase.com/timotheus