photo editing programs

[David Luttmann]

Good description Marko.

The primary reason for doing work in LAB mode is to avoid applying certain adjustments to the 2 color channels. This is the best routine to take....but not in 8 bit mode. Converting to LAB mode in 16 bit is fine. While still destructive, the change in color and luminosity gamuts and tonality will be moot in 16 bit mode as there is more than sufficient data to avoid problems.

Converting to LAB is better than just using the Edit/Fade/Luminosity commands. USM then using fade luminosity produces DIFFERENT results than sharpening the L channel in LAB mode with the same amount of USM.

The reason is that L channel sharpening will only detect edges of different luminosity. Sharpening the colour image then fading luminosity will prevent colour shifts BUT sharpening will have occured along edges with different colours even if they are the same luminosity. The difference is minor, but it can be detectable in print.

Doing this in 8 bit mode is just asking for problems.

[Jonathan_6488]

Well - sorry if I just happened to have a pickle a little but up my hiney while writing that. But my goodness... such responses...!

Let me address these, if I may;

"1) No such thing as NTSC JPG format. JPG is JPG."

The last time I checked - there was. It is my belief (attack me if I'm wrong here, he said sarcastically) - that JPEG files can be different resolutions. Do you not agree with this?
NTSC is a format established by the National Television Standards Committee which refers to a video image displayed on a television or computer monitor which is exactly 480 pixels TALL and 640 pixels wide. Please google it if you don't believe me. It has absolutely nothing whatever to do with the file storage protocol, which in this case, is JPEG. If you've done any work whatever with image editing in the professional world, or know what you're doing or understand the history of image processing on the computer, then you would probably immediately recognize NTSC as referring to a resolution spec. It means, flippantly, LOW res.

"2) Editing in LAB is a destructive conversion process. In 8 bit mode, you will loss approx 10% of your tonal information on the conversion from RGB to LAB & back. "

Not if you can start with LAB images. Which IS the case with some software and digital capture methods. But I fullyagree if you're starting out with RGB images.

3) NO authority on printing or workflow would support your arguments in favor of 8 bit processing being equal. It is ALWAYS better to make ANY adjustments in 16 bit mode in order to preserve all the subtle information. Any changes you make are cumulative in nature....not just changes to density.

Where are you going to get Photoshop CS2 for less than $100? Was this not the original point? I was simply trying to make the argument that 8 bit/channel image editing software. 16 bit IP is not necessary at all. I maintain that it is completely possible to have an 8 bit/pixel image that is 'adjusted' in 8 bit/pixel color space which is colorimetrically and densitometrically QUITE acceptable when compared to an identical image that is corrected in 16 bit/pixel color space. PROVIDED THE OPERATOR KNOWS WHAT THEYARE DOING. Again - please read what I actually wrote and not try to draw your own inferences from buzzwords. The suggestion I was trying to make was that it would be CHEAPER but harder. And therefore the user would verylikely, in my opinion, learn more.

"4) Film is 1 bit in nature with dithering???? I won't even bother addressing this ridiculous statement. "

If you can't see how I could make this (contentious, of course) statement... well I won't say any more. Hint: please don't read it TOO literally.

"5) Assuming you don't have a 3Ghz Dell with 64mb of RAM, there is NO WAY your Dell will choke on a 30mb file. A 3Ghz Dell with a gig of ram will run circles around your old G3 with the same memory. "

Okay - well, guess I'm lying. Would you care to put some money on it? Mypoint was that processor clock speed isn't anywhere nearly as important as having a wholly optimised system, especially, perhaps, disk subsystem. There are other factors that will bottleneck you much more quickly.

"6) You old Photoshop versions don't support a lot of the 16 bit functions that are necessary to maintain a decent image. "

I would have to assume that you either have no idea what you're doing (and haven't used the program much) or else have been listening indiscriminately to painfully unintelligent salesmen. Would you like to have a shootout sometime (hopefully NOT with guns, please!!)? We could have a third party prep an image for us both to correct and see who gives back a sweeter histogram...? Given your replies, I'm not too worried about the competition. Sorry - but that's just the way it is.

"7) Yes, I'd say anyone doing fine art printing in 8 bit mode has comprimised their work. I have yet to see high quality printing examples that supported your views that 8 bit & 16 bit processing routines are equal. "

Well, I'm talking about mostly magazine work, etc. here... but I really cannot be bothered to supply anyone with examples. Perhaps this is best addressed by the previous issue (no. 6).

"8) Lightjet or web does not need to display more than 8 bit. But by starting in 8 bit mode, and converting to LAB, and making adjustments from there, you are no longer left with the full 8 bits....you have lost tonal information. Thus, starting out with 16, making adustments that leave you with 14 for example, means you still have a clean amount to convert to 8 bit. This isn't difficult to understand. "

I absolutely agree with you on this point. In theory. I see nothing wrong with what you're saying. I think that would be a riduculous way to do things. I would only edit LAB if I were able to start there.

"9) On average, the healing brush is far easier to use than the cloning tool for removing blemishes on your scans...ie; sky areas. "

Well - I'm more comfortable with the cloning tool. Call me old fashioned. But for smooth areas - yes, it makes sense.

"So Jonathan, don't get upset with everyone for pointing out your mistakes. Unfortunately, your help is actually a hinderance when you throw around processing comments that in fact are just not true. "

Hopefully you now understand what I'm trying to say - that we can come to more of a mutual agreement here - and that you will make more of an effort to understand what you're attacking before you make such ignorant blanket statements. Thus far we've learned that a few of these items were false assumptions. And I believe you stand corrected on the points that you'd taken as the greatest of my perceived errors. I'm HOPING that you can see some sense in the point I was trying to make and the parameters within which I was making them (again, and again - please please see the original post and don't take peoples' comments out of context).

Sincerely,
Jonathan

[Jonathan_6488]

Marko,
Your contribution is appreciated. But my last response to the other poster also - well, it must apply also to you. Please read it. How on earth is film NOT like one-bit with dithering?? Should I qualify it to help you understand this better - by tacking on the phrase "in a random multilayer bit field"? You have a hit or you don't as far as the silver coupled grains are concerned. Simple as that. ONE BIT. Three or four color layers - one bit per layer. Right? Get it? Anyway - I wasn't trying to suggest one was an exact analogue for the other. I was trying to make a fanciful statement that would get one thinking about the bigger picture.

So anyway - the point you were trying to make - that 'cheap' and image editing being necessarily expensive - well, that's just not even logical. You read too many magazines. So you say that CS2 is the cat's pyjamas...? But in three years, according to your criteria - it will be an absolute piece of crap. That makes no sense. I smell bad smells - you're being duped. A pixel is a pixel. I don't think the file knows how expensive the machine attached to the processor was. Technique is everything in my book.

J

[Jonathan_6488]

Dave,
I was talking only about color correction in LAB mode, assuming the source file is in LAB. Not RGB. But thanks for your input. You're obviously a bright guy.

[David Luttmann]

Jonathan,

1) There is NO NTSC JPG format. Get off this....please.

2) What scanner or digital capture device are you using that starts with LAB? Or finishes in LAB?

3) Knowing what you are doing does not change the fact that any changes you make in 8 bit mode will reduce your tonal accuracy. This is not an issue in 16 bit mode. Thus, you can change NOTHING in 8 bit mode without reducing the quality of the image. Sorry, but the value of the software is irrelevant. Editing in 8 bit mode is ALWAYS bad....period!

4) We'll leave this 1 bit comment aside.

5) Didn't say you were lying...just that your a bit over the top with this ancient G3 vs a 3 ghz Dell, equal in memory.

6) I"ve been using Photoshop since the beginning. Try using all the filters, etc, in 16 bit mode in version 4. Oops, you can't, because version 4 locks out 16 bit processing on a lot of filters. I've never used a salesman for purchasing software. But I'm afraid you're fighting an uphill battle. Because you don't seem to think processing in 16 bit is necessary, I'm not suprised that you'd think version 4 is fine. Unfortunately, you're alone in this thought.

7) Magazine & web work is low quality. I'm talking about maintaining the best quality....not going after the lowest common denominator.

Jonanthan, you're problem is that you refuse to accept some very basic processing truths. Even if you could start in LAB....and not many routines allow you to start this way....any changes you make to the file wil result in less than 8 bits of data. These changes include:

- Any downsampling or upsampling to achieve final print size.

- Any dodging and burning done on the image (virtually ALL require some)

- Sharpening

- Adjustments to levels, contrast, saturation.

- Any of the above changes made in order to accomodate different papers, printers, inks, etc.

In the end, you cannot use 8 bit in LAB to achieve the same result as 16 bit. As well, it is not recommended to print directly from LAB to an inkjet...and Lightjet does not work with LAB the same as RGB. Thus, you MUST convert from LAB to greyscale or RGB at some point. So not only will you loose information here, but you lost it to begin with by starting your workflow in 8 bit...no matter how much you know.

Sorry, but no matter what you are doing, it is ALWAYS better to do it in 16 bit mode first. If you're doing magazine work or web work, I would still do it in 16 bit mode. Why are you willing to chance an impact to the quality of the output of your work?

Regards,

[Frank Petronio]

I don't feel compelled to redo images I created back in 1998... they are fine, even though I had to use only 8-bit, etc.

The advantage of using CS2 is that I can get to where I want to be in 4-5 steps rather than 20, and end up with a fine base image with more "overhead" for further editing should I need to get there.

If I wanted to redo my 1998 images, I now would be able to work on larger scans faster, and probably get smoother transistions in difficult areas like skies and soft, smooth sections.

[Marko]

Johnatan,

There is a fine line between being persistent and being headstrong.

NTSC is a format established by the National Television Standards Committee which refers to a video image displayed on a television or computer monitor which is exactly 480 pixels TALL and 640 pixels wide. Please google it if you don't believe me. It has absolutely nothing whatever to do with the file storage protocol, which in this case, is JPEG. If you've done any work whatever with image editing in the professional world, or know what you're doing or understand the history of image processing on the computer, then you would probably immediately recognize NTSC as referring to a resolution spec. It means, flippantly, LOW res.

OK, now you have me confused - you're first calling NTSC a "format" but then you change that to a "resolution". So which one is it, a format or a resolution?

As for JPG being a "file storage protocol", whatever that meant, I won't even comment ont that, see below.

Now, as for "image editing in the professional world", I'm even more confused here. I've been working for various magazines and/or online publications in one role or another for longer than I really care to remember, I've handled many different image (and file) formats in the process and never once did I come across the need to work with either NTSC (in the U.S.) nor PAL (over in Europe). Much less did anybody I know refer to them as "image formats"

NTSC (and PAL) are video image formats. They do not specify resolution in pixels, they specify resolution in the number of horizontal lines. And they also specify the "refresh rate", or the number of times half of the resolution lines (every other line) is drawn in one second. Without getting into real technical specifics, like carrier signals, let's just say that NTSC and PAL differ in both the number of lines and refresh rates.

640 px x 480 px, on the other hand, is a pixel resolution for a still image. It is commonly known as "VGA" resolution. I have a feeling that this is what you actually meant by "NTSC resolution".

It can be stored in a variety of "image formats", with or without compression. If a compression is used, than it can either be lossy or lossless. JPG is one of those formats (recently two), and it employs lossy compression of varying levels.

Finally, file formats have nothing to do with either images or any other type of document, they determine how a flow of data, any data, is physically stored on a permanent storage in a way that could be reliably retrieved back into working memory, This depends on a computer system and has nothing to do with either NTSC or PAL. Or JPG for that matter.

Should I qualify it to help you understand this better - by tacking on the phrase "in a random multilayer bit field"? You have a hit or you don't as far as the silver coupled grains are concerned. Simple as that. ONE BIT. Three or four color layers - one bit per layer. Right? Get it?

Sorry, I am not impressed at all with fancy phraseology. Empty one at that. For starters, a bit is either a mathematical entity and as such does not have physical size at all (making impossible for light to hit one) or they should all be equal size and shape, as in a half-tone raster, so that the packaging density determines the shade. Silver particles are anything but either of these two models. They are random in both shape and size and density. Also, there are no silver particles in color layers, as far as I know.

So, I get it, allright, but the question is: do you?

[Jonathan_6488]

Marko - you're totally exhausting me. But whatever. Think what you want to think. I am sorry if I'm using a somewhat antiquated term. NTSC 'format' does not get used much these days. However - if you look on the image res settings of popular consumer format digital cameras these days you will also see a holdover from this sort of 'lingo' (VGA for example)... though I don't think it's 'fancy' just an easy way of saying 480 x 640. So - sorry if that's a bit obscure. I just figured a lot of people here would be old school enough to appreciate that.

BTW - the 'fancy phraseology' , if I'm understanding you correctly - is not meant to make unclear or to try to 'impress' you. I am really kind of depressed that you might think that. I am trying to be CAREFUL with my wording so that you get what I am trying to put across to you and that I might eliminate any further misunderstandings.

Anyway - as for the layer thing.... well - I would CONTEND That many color film manufacturers refer to their film as having a layered structure. The "dye layers" or "dye-coupled layers" are in fact silver grains which, when reduced by the developer produce density of the color relevant to it's particular layer. This is the way that color film has always been made. If you can prove me wrong, I'm happy to eat my words. But to the best of my knowledge, this has always been the case. I really hope you don't want me to have to prove this to you. Ihope you will bother to take the time to find this out on your own. It's something you should know about - sounds to me.

As for the 'one bit' thing... dare I further condescend to explain how this works?? Guess I have no choice.

one bit = two possible states (on/off) e.g. 1 or 0
two bits = four possible states - e.g. 01 or 00 or 10 or 11.
three bits= eight possible states - e.g. 000, 001, 011, 111, 010, 110, 101, 100.
four bits = sixteen possible states - e.g. 0001, 0011, 0111, 1111, 0010, etc. etc..
five bits = 32 possible states
six bits = 64 possible states
seven bits = 128 possible states
eight bits = 256 possible states...

still with me ?? good.
So - silver halide TYPICALLY have only TWO states. on or off. Hit with enough energy to trigger a reduction reaction or NOT hit with enough energy to trigger a reduction reaction.

So - it would seem to me - according to this model that I am using to conceptualize the photographic process that a silver halide grain has behaviour NOT UNLIKE a one bit word in digital parlance. If you still don't get the point I was trying to make -then, well...whatever I guess. We cannot all be on the same page all the time I guess.

Anyway -I'm HOPING that my original thoughts are just a bit clearer to you now.

Thanks for bearing with me - and good night.
Jonathan

[Jonathan_6488]

Dave - please think of it this way - sub out the word NTSC for 480x640 and see if that works better for you. Or would you still refuse to believe that it exists...? If you want I can send you a JPEG of this 'format' (yes, my loose words).

As for the whole 16 bit thing. Yes - I'm agreeing. yes,it would be more desireable. I never SAID it wasn't. We're having TWO entirely DIFFERENT discussions here. The Brotha just wants to get some CHEAP software. I'm saying older photoshop versions AIN'T that bad. You can do really reasonable work on them.

that's all.

[vamp007]

Well, the last post in this thread is more than a month old, but I believe the "Pixel" editor can be interesting for all of you. It's available at http://www.kanzelsberger.com/pixel/ and has many interesting features for a very reasonable price ($32 at this time). For example it runs on many platforms (Linux, Windows and MacOS to name the most important ones) - I use linux all the time, and for me it has several advantages when compared to GIMP - 16bit support, CMYK etc.

It's not in the 'stable' version yet (but it'll be available very very soon), but seems very stable to me. I definitely don't need photoshop when I have this. But remember - all of thsese are just my personal opinions, yours can be completely different.