TIFF vs JEPG
in [Scanners]
|
Prev: canon F914900
Next: Canon FB 630 U - Driver
From: Don on 21 Apr 2006 14:12 On 20 Apr 2006 08:16:04 -0700, "Noons" <wizofoz2k(a)yahoo.com.au> wrote: >> That's one example. Another one is areas with lots of high contrast >> detail where the 8x8 pixel JPG squares are as plain as can be. > >Bingo! And why I make a point of NEVER commenting >on a "crop" of an image without having the original >ENTIRE file to compare against: it's only too easy to - intentionaly >or not - make a crop at anything other than 100% where artifacts >don't show as well or get masked by other noise effects. >With the original file that is a lot harder to do and it becomes >clear which is jpg or tiff. Exactly! Which is why establishing unbiased testing environment is crucial if the test results are to have any meaning. And also the first step in any objective methodology. Our worthy interlocutors, however, are too busy hurling insults to even notice that, let alone even make an attempt at a level playing field. With such entrenched feelings trying to reason with them, at this point, is futile. >> Both of those are statements of subjective opinion (i.e. a reflection >> of how much one values image content). I never talk about that because >> that not only depends on the context but each person has their own >> individual requirements. I only talk about objective facts and, as you >> yourself confirm, clearly visible JPG artifacts at 100% are a >> self-evident, axiomatic fact. > >Couldn't agree more. Unfortunately, others have made up their minds long time ago (from the start) and are now just spoiling for a fight. My refusal to enter into pointless name calling and abuse just seems to make them angrier. I do bend over backwards in order to be as objective and as fair as can be. But that should not be confused with bending in the other direction, as some here would have me do! ;o) No chance of that, I'm afraid. Don.
From: Don on 22 Apr 2006 10:35 On Fri, 21 Apr 2006 09:09:53 -0400, Bruce <XXbg2gX(a)XvirginiaX.Xedu> wrote: >Here you are accusing me of not >having read the thread, which is an unfounded insult. No, an objective fact without even a hint of a personal attack, as will be shown in the following paragraph. >>> Yes, everyone agrees that with high compression JPG artifacts can be >>> objectionable. But the point is that at low compression the quality can >>> be so good that any data loss is imperceptible even after zooming in. >> >> That was never the subject, Bruce. It's an unrelated tangent. Please >> read the messages. > >Here is a statement about tiff vs jepg (which is the subject line of the >thread): > >> If you blow up the images to the point where you can see >>>individual pixels, you will see that some of the pixels are >>>different in the two images. But even then, unless you know >>>in advance, you won't be able to tell which was the original, > >And here is a direct quote from you attacking it: > >> That's just patently false! >> >> At 100% magnification (i.e. 1:1) even a JPG image at lowest >> compression (i.e. highest quality) stands out like a sore thumb when >> compared to the original. Now then, what does that *objective* statement of fact have to do with your *subjective* feelings above (on top) about individual estimation of JPG artifacts ("objectionable", "good", etc)? I've started out (and repeated it often enough, last time in my reply to you!) by stating that *subjective* feelings (what is good, what is objectionable) is *not* the subject and, indeed, said words to the effect "to each his own" which Alan acknowledged (!) but many others seem to miss as they trip over themselves to hurl insults and abuse. So, talking about *subjective* perception i.e. how much is "too much" or what "looks good" is *not* the subject! Which is why I (calmly and *politely*!) merely suggested reading the thread again. No insults! >Numerous people have posted examples to show that this is not >necessarily the case for high quality jepgs. Your response has been to >question the "veracity" of their examples. No, again that's just factually incorrect. For starters, there's nothing there to even test the veracity of!! You're parachuting into the middle of the thread and misreading a single word, drawing totally unsubstatiated conclusions from that and thereby missing the context completely. Which, again, is why I (calmly and politely!) suggested reading the whole thread again and taking the full context into account. From that, in a nutshell, my response was: "First we have to establish an objective testing environment" and then proceeded to outline some parameters to cover the basics. Veracity of data (from *both* ends!) is an *essential* part of objective testing! As is something called "can't be falsified"! STOP! Don't jump to wrong conclusions now as you have re veracity! Please read up about this, first! This "can't be falsified" is *not* an accusation but a most fundamental aspect of objective testing. It protects *both* sides from *inadvertent* errors and assumptions. (In science, one way to verify a theory is by proving it can't be falsified.) Anyway, the response to my trying to define an objective testing environment was insults and abuse (repeat: which I did *not* return!!!) and posting of so-called "examples" which violate virtually all conditions for an objective test showing they either did not read or did not understand the thread. To preempt your reply, that's not an insult but a fact! If they read or understood the thread they wouldn't have posted useless "examples". >Well my own experience is that you are wrong. If you are interested is >"truth" and "facts" post examples to educate us and show us conditions >where a single high quality jepg compression does cause visible image >degradation. As I already indicated, at this point, that's a total waste of time for a number of reasons e.g.: - If people start the discussion by hurling insults (without reading) what hope is there they will now all of sudden become reasonable? - If people can't even grasp the most elementary parameters for an objective test (e.g. rush with unsuitable "examples" violating and ignoring explicit instructions) what hope is there they will all of a sudden become thorough and reliable? Especially in their current overagitated state? And, no, that's not an insult, but a fact. How else do you explain that after I went through all the trouble of outlining the basic parameters the response was to totally ignore them (not even challenge them!) but just go ahead and post totally unsuitable "examples" violating all of the most elementary principles of objective testing? So, the only prudent course of action at this point is to ignore the insults and abuse (so as not to perpetuate a useless thread) and focus on the facts which is I have been doing and what I'm doing here. I've given enough information for anyone who wants to perform the tests themselves and will be happy to provide more. Therefore, a careful reader can take it from there. >My understanding is that the jepg standard uses several separate >compression schemes most of which are disabled at the higher quality >settings, but each software package implements the details differently, >hence the need for actual testing. Without getting into details there are two key aspects: one, the original JPG specs are very "flexible" (no set quality scale) and two, there are no provisions for lossless compression. That's all we *need* to know in the current context. Knowing how JPG works (e.g. the 8x8 pixel squares) is helpful, but not really essential. Don.
From: Raphael Bustin on 22 Apr 2006 11:33 Don, you're such an shameless, clueless gasbag. Instead of tiliting at strawmen or opining aimlessly about science and methodology, please just show us ***one example*** to prove whatever point you're making about JPG. Your words have lost all credibility. rafe b www.terrapinphoto.com
From: Scott W on 22 Apr 2006 13:34 Don wrote: > So, the only prudent course of action at this point is to ignore the > insults and abuse (so as not to perpetuate a useless thread) and focus > on the facts which is I have been doing and what I'm doing here. I've > given enough information for anyone who wants to perform the tests > themselves and will be happy to provide more. Therefore, a careful > reader can take it from there. Come on Don, you said the jpeg would "stick out like a sore thumb". When shown examples showing that you can't tell the difference you have said they were not valid. Why can't you post one example? Scott
From: Alan Meyer on 22 Apr 2006 18:04
"Alan Meyer" <ameyer2(a)yahoo.com> wrote in message news:uJKdnfHCnLcuoaDZnZ2dnUVZ_vmdnZ2d(a)comcast.com... > ... > I believe that the TIFF vs JPEG issue is a red herring. I defy > you to look at an image on screen or on paper from a TIFF > vs. one from a JPEG compressed to 1/10 original size, and > tell which is which. > > If you blow up the images to the point where you can see > individual pixels, you will see that some of the pixels are > different in the two images. But even then, unless you know > in advance, you won't be able to tell which was the original, > and you still won't be able to see the differences when viewed > as images instead of as individual pixels. .... After all the controversy I went back and looked at the images I've been making and saving - both from my scanner and from my digital camera. What follows are my __subjective__ impressions. I'm not claiming that everyone else will perceive things the way I do, but I bet a lot of people will see the same things. My normal technique is to save using JPEG at compression levels of between 10:1 and 20:1. I arrived at those numbers by trying numerous different compression levels from 3:1 up to 100:1, on a number of different images from high detail to low detail subjects, and comparing the images. The actual compression I use depends on settings in my scanner, my camera, or in the GIMP, which produce results that I liked. The output of these settings is typically between 10 and 20 to 1, depending on the detail found in the image. Going back after all of the discussion, I wanted to see if in fact there were artifacts that I had missed. Here's what I found: If I magnified my images 4 times, (double each dimension), I could often see JPEG digital artifacts - i.e., compression squares. Pixelation of diagonal lines also became visible, but that's a digital artifact, not a specifically JPEG artifact. Other JPEG artifacts, such as the clumping of colors in expanses of small color change areas such as the sky, were sometimes visible but usually not. Then I reduced the images back to original size, keeping my eyes focussed on the JPEG square borders. To __my eye__, the JPEG artifacts disappeared. Back at 1:1 resolution, try as I might, I could not see the squares. It has been asked, "Don't you ever crop?" The answer is, yes, I do. But if I need to crop and blow up an image, I do that by scanning at a higher resolution, not be reducing JPEG compression. As someone said earlier (Scott perhaps?) for any given number of bytes, we'll always get a better image out of JPEG than TIFF. If I need to blow up a small part of an image, I can always get better results by scanning to more pixels, then compressing with JPEG, than by scanning with fewer pixels and saving in lossless compression. The non-compression related artifacts such as diagonal line pixelation will always look better in the higher resolution scan. So, I will go even further than some of the JPEG defenders have gone in this thread and say that some of us don't even require the highest quality JPEGs. For some of us (I suspect for many of us) something less than that will work just as well. Finally, I will reiterate that my practices aren't right for every viewer or every application. Some people want and/or need higher quality images and don't care about the extra storage. Many people scan at higher resolutions than I do, and save images with less or no lossy compression. However, I bet that at least some of the people that insist on these high quality images are kidding themselves. This whole debate reminds me of a time almost 40 years ago when I was working as photographer for my college newspaper. I was using a Yashica rangefinder 35 mm camera that I had bought for a few bucks in a pawn shop. It was all I could afford. Another fellow, from a wealthier family I guess, had a Leica. I acknowledged that his Leica would take sharper images than my Yashica, but said that for many subjects and shooting conditions it would be very hard to tell the images apart. He became very indignant and challenged me to a test. He pointed his camera at something and said let's both shoot this image. I said, "There's not enough light, we'll be shooting at 1/15 second shutter speed and get blur from not holding the cameras steady." He replied, "I can hold the camera steady at 1/15 second!" (Anyone who has studied photography knows that is false.) It's easy to get wrapped around the axle on a few aspects of image quality and lose sight of the big picture of what you're really trying to do. Alan |