eric_perlberg

Ben, what colour space are you using when converting your RAW file into photoshop?

You don't say how you're prepping the file before using Save For Web. If you first convert the images to sRGB before using Save For Web you'll find less loss of saturation if you're not already doing that. Save for Web strips away any embedded profiles so you have to use convert to get and keep them in sRGB.

You don't say where you are but in London you might want to look into Photofusion courses (can't say if they're good or not though they seem to have serious kit): http://www.photofusion.org/education/courses.htm

I also apologise to the original poster but since we're into this this far might as well see it

through.

 

Mac use of colour management is notorious for not being joined up, for example the

teams that work on these various bits like colorsync itself and then Preview and then Safari

don't necessarily share the same fundamental concepts.

 

What you've established, and I concur, is that there is some anomoly in the way Safari

interprets data with and without a profile is not accurate. So be it. Now that you bring it up

I remember reading exactly your statement and this strange behaviour on the colorsync

list. It's a tiny little blip in Apple's implementation or misimplementation of its own

standards.

 

Now lets go back to the original post and your original answer. This chap Greg asks what

he's doing wrong as his images for the web don't look right. You tell him to convert to the

monitor profile. Given all that you and I just went through and your final statement that

Safari has some glitch in interpreting data with and without an embedded profile, why did

you tell him to convert to his monitor profile?

It seems to me you've answered your own question. The image on the left if I understand

correctly is converted to sRGB and in the Save For Web dialogue its saved as a jpg with no

embedded profile. The image on the right is done the same way, convert the data to sRGB

and in the Save for Web dialogue you attach/tag/embed your monitor profile. The data in

the file is the converted sRGB data. There is no longer the aRGB data in the file.

 

But by telling safari to interpret the sRGB data in terms of your monitor profile, Safari

reinterprets the sRGB data and comes up with the image on the right. This shows the

power of your monitor profile on both the aRGB data in photoshop and the Safari jpg with

the embedded monitor profile. While it may seem that embedded profile jpg is just

remaining more accurate, I don't think so. The sRGB data in the embedded profile jpg is

being stretched to fill your monitor colour space. Doing this in Firefox would result in a

very different set of results. The same for IE on a PC but IE on a mac will work similarly to

Safari.

 

That this file with the embedded profile is to you more "accurate" is the result of two other

situations, the first is your monitor colour space carried in the embedded profile and the

second is that Safari is a rare exception of a web browser which can interpret data in terms

of an embedded profile. If you open the files in Preview you'll probably have the same

experience because it too can interpret data with a tagged

profile. Import both images into Word and you'll see something very different.

 

The main point being that (I"m assuming I'm right, its possible to know what one is talking

about and still be wrong...) in your last set of screen images, both files "only" have the

sRGB data to work with. In the latter, that data is "stretched" to fit your monitor profile and

results in what in you've shown us. Again, if you printed both the original aRGB photo at

reasonable size (say 8x10) and you printed the sRGB file with your embedded monitor

profile at the same size, I bet you would see a difference between those two assuming it

was a printer capable of showing aRGB colour space.

Maybe this example will help.

 

Say you have an aRGB file with the numbers 1 through 20. Your monitor is capable of

showing the numbers 1 through 15 and sRGB is capable of showing the numbers

1 through 13.

 

When you're looking at your aRGB file on your monitor you see the 1 through 15

data in your file but you don't see 16 through 20 because your monitor can't

display that. Whether you convert this doc to your monitor profile or embed/tag the

monitor profile to the doc, from Safari's point of view it will show the numbers 1

through 15 and voila it will look just like your photoshop file did.

 

Convert to sRGB and you'll see the numbers 1 through 13 and you're saying, hey!,

my monitor profile makes my image look just like photoshop and its closer than sRGB.

 

I think that's what's going on.

<i>I think the issue is that Safari applies my monitor profile to untagged jpegs.</i><p>

Not exactly<p>

When you <i>convert</i> your doc to sRGB and then look at it in Safari, there is no

profile attached (ie, tagged, ie, embedded). Rather the data in the jpg you're viewing only

contains sRGB data (the rest having been thrown away in the conversion process, so its as

if it were tagged in sRGB but there are no tags involved.<p>

If you're using <i>convert to profile</i> on the second jpg, then there's not a monitor

profile attached to this doc either, the info has been converted to and is limited to the

monitor profile space data<p>

What Safari is capable of doing but which you're not doing in either of your examples is

reading "tagged" or sometimes referred to as "embedded" profiles. Here all the original file

info is kept (in your case all the aRGB data) but its tagged with a different colour space

marker which Safari is able to understand and then interprets the aRGB data (all of which is

still with the file) and displays it as though it were converted to sRGB or your tagged

monitor profile or some other tagged colour space.<p>

I think the sticking point is that "tagged and "embedded" mean all the original data is still

available but a tag is attached to the file tells a colour aware programme how to interpret

the data (you're not doing that) whereas "converted" means there are no tags, instead

some data is actually thrown away in the conversion process.<p>

But as I said in the post above, the reason that your conversion to your monitor profile

looks identical to your photoshop doc is because essentially it is identical as far as your

monitor can show you. This implies that your monitor is capable of displaying a slightly

larger colour space than sRGB which seems plausable but that info would be useful to

know to test my hypothesis.

I'm not aware of emacs being difficult to make monitor profiles for, at least I haven't come

across that problem. Do you know how bright you had to set it in cd/m2?

<p>

Here's what I think is going on in your screen shots.

<p>

When you look at a photoshop doc on your monitor and its in aRGB, the monitor is doing

its best to show you how the document would look in aRGB within your profiled monitor's

limitations of colour gamut. Roughly speaking your monitor is about at the level of sRGB,

maybe a tad better.<p>

 

When you convert to sRGB as in screen grabs one, there will be a change in certain areas,

especially more saturated colours and when you view this converted to sRGB image in

Safari you're seeing the slight loss of saturated colours, esp in the blues in the sky and

maybe a bit of the greens and some red tint in the ground.

<p>

<a href="http://www.drycreekphoto.com/Learn/color_spaces.htm">Here</a> is a

comparison of the two colour spaces and as you can see at the bottom, the blues in sRGB

are clipped compared to their aRGB cousins. Now the diagramme at this web site is flat but

colour spaces are really 3D (the dimension missing here are the gradations towards pastel

and eventually white)<p>

To further complicate matters, you can't see any colours that your monitor can't produce

whether or not they are in sRGB or aRGB. So when you do your aRGB to sRGB conversion

you are seeing the loss of saturation mentioned above but the photoshop aRGB document

is already limited by the monitor profile so that's the extent of your emac's ability to show

the aRGB doc regardless of what's really in it.<p>

But in the second screen grab you've converted the aRGB doc (which probably contains

saturated colours than you can't see on your monitor because its limited by your monitor

+profile) into your monitor profile which was the limiting factor in any case, the result is

that you're not seeing any difference. However, I'll bet that if you have a good printer and

printed out all these files a) you'd see a difference between the sRGB and aRGB doc on the

one hand AND b) you'd see a difference between the aRGB and your converted-to-monitor

image on the other. Does that make sense?

Hi Eliot,

Is the saved for web jpeg in Safari in the screen dump above the one that's been converted to

your monitor profile or is it in sRGB?

<i>as whilst the JPEGs may look a bit out of whack on your own computer, by and large they

will look ok on the average computer.</i>

<p>

I don't find that converted images (aRGB to sRGB) look out of wack on my computer screen.

Since Greg's original query was about "JPGs as I'm saving them for web use" I would have

thought he was interested in showing the images to others not looking at his monitor?

I don't understand the logic of converting to one's own monitor profile. Since everyone has

a different monitor and the profile is specific to your own, this only guarentees that your

docs look good on your own monitor. There's a certain amount of variabiliity you have to

accept with images on the web but it shouldn't be as bad as your samples.

 

Any professional that I've ever spoke to and my own rather extensive experience creating

images for the web dictates to convert to sRGB which is about as close to a generic

monitor profile as there is.

 

Web and print images need to be prepared separately unless you're sending images to a

printing service which requires or is based on sRGB as the "profile". Un-colour managed

programmes are simply passing through the colour numbers which they are sent. Since

sRGB is the lowest common denominator in the colour management world, its the best for

these situations to have the file converted so that the raw numbers passed through are

sRGB numbers.

 

I don't see the kinds of shifts you're getting in your un-colour managed screen dumps

unless I fail to convert to sRGB and the doc is in aRGB or prophoto. Could you have a

currupted (or wrongly named) sRGB profile in photoshop?

There was just an analysis of the cost per page of printing on the 2400 at the Epson 2400 yahoo group. Go to Yahoo and then select Groups and search for Epson 2400. Also some interesting analysis of the cost of doing various head cleanings and chnaging black cartridges.

 

The 2400 is a very nice printer IMO.

No the display shouldn't be pink (you don't by chance have a 23" Apple Display? if so, many have this pink problem... if you can get your screen looking neutral you should be relieved that you can). Choose native white point rather than 6500 even though it won't be very different. Since native white point makes your monitor almost perfectly neutral its definately the better choice. Don't know why Match 2's easy mode defaults to gamma 1.8 but your Apple LCD has a native gamma of 2.2 and you should choose that. The old addage that macs use 1.8 gamma is left over from earlier days and you can safely ignore it.

I have to admit that I'm on the edge of my knowledge/ignorance here but from what I

understand, the perceptual intent comment which shows up in the colorsync dialogue box

relates to the ICC's specification of how to transform any input colour into "a well defined

virtual colour space" called the Profile Connection Space from which an output profile can

be built. All ICC profiles by definiton use Perceptual Intent at this initial transformation

stage where the input colour is brought into the PCS. The PCS, based on the CIE 1931

standard colorimetric observer, is necessary to create an unambiguous and clearly defined

starting space from which all output profiles can then be created in such a way that they

can be decoupled from the equipment they're produce with.

 

What you are chosing in Rendering Intent when you make a choice in Print

with Preview is a second transformation which goes from the PCS to the printer. This is not

predetermined by the ICC and relative colorimetric is more accurate for colours in the

source space which fall inside the gamut of the output space whereas Perceptual Intent

will produce a seemingly better match if there are significant colours in the input space

which lie outside the gamut of the output space.

 

I don't believe that the implication that Perceptual intent is the native profile definition and

therefore the first and naturally best choice in using a color conversion engine in actually

using the printer profile. It's my understanding that this is a separate function of the color

conversion engine. Otherwise, why even have relative colorimetric? More information on

this topic at www.color.org/profile.html.

 

Although I can't validate this from my personal knowledge of Photoshop programming, I've

read and hopefully understood correctly from the likes of Andrew Rodney and Bruce Fraser

who are beta testers for photoshop, that in fact, in photoshop, internal colour space

conversions (but not printer colour conversions) are only done in relative colorimetric even

if one chooses the perceptual intent option. The only place that Perceptual Intent is

actually available and functions is in the Print dialogue box.

 

If I'm wrong on either of the above I'm happy to learn the correct info.

Yes, keep the original in Adobe RGB for future printing. Yes, sRGB is currently your best bet

for getting something approaching consistent on monitor viewing.

when i was testing out to see if i could get by with an iMac I checked with file sizes at

25mb

(6-8mp camera) and with 95mb photoshop files (my 1dsmkii). I made multilayer docs and

working in CS I applied various filters (it was at an Apple Store and they had no plugins

beyond straight CS). I thought the 95mb files with layers doing sharpening and bluring

was

not really desireable but doable (I know professional photographers who do some

processing on PowerBooks with a G4 process.or, it can be done but not fun. Curves and

Levels were less of a problem. Didn't check noise

removal but I would guess with 50meg files you're looking at 60-80 seconds. I think

50meg

files are borderline, but definately max out RAM. So much depends on your workflow and

expectations. There's a page somewhere I can find if you want which goes through various

Photoshop strategies to get the most out of what you have.

For digital images from 5-8mpx cameras working in 16bit the iMac should be fine (as I remember the files are about 25mb in size in photoshop). If you buy your dream 5D, you might begin to have visions of a dual desktop. RAM from <a href="http://www.crucial.com">Crucial</a> is high quality and much cheaper than buying from Apple. If you can, bring the machine up to its 2gig max. Also consider backup. You can use the firewire to get a backup HD or less robustly back up to DVD. <a href="http://www.shirt-pocket.com/"> SuperDuper</a> is cheap, easy to use and widely respected backup/recovery programme for macs. Enjoy!

Obviously comparing a known camera to an unreleased and unreviewed camera is guess work. I assume that the 5D will have better low light characteristics (ie, less noisy when shot at 800-3200). Indeed the 1Ds only goes to ISO 1250 if memory serves correctly.

 

I do a lot of handheld urban architecture photography and find that my 1DsII is not too heavy to walk with but a) I use the hand strap a lot and b) I don't ever carry more than 2 lenses and usually just 1. It isn't light by any means. The 5D trims a pound or so off that weight. Another question which may affect your decision is battery time. I don't know about the 5d battery but the 1Ds I remember not being nearly as good as the 1DsII. dpreview may have some details on that.

Ron, from your description of the "soft-proof" window, you're not using soft-proofing, you're actually using the Print with Preview view of your image. Soft-proofing uses the same window your regular work window uses and is toggled. This may be part of your problem as Print with Preview is only meant to show you location. <a href="http://computer-darkroom.com/softproof/softproof_1.htm">Here</a> are instructions for soft-proofing. I think you'll find window size is not an issue.<p>

As you're happy with the colour accuracy, I wouldn't have thought that choosing a rendering intent (perceptual or relative colorometric) would affect your images to be darker or lighter than your soft-proof (but as above, so far you haven't really soft-proofed if I understand you correctly). That being said, choosing between the two requires a bit more sophistication than above. It really depends on the colours in the image and is well explained I think in your books. Basically you use relative colorometric as your first choice but if the image has a lot of out of gamut colours (you can use the gamut warning just below the soft-proof choice in the VIEW menu) in which case Perceptual will give a more accurate print. The two are different ways of mapping colours from your document colour space (Adobe RGB in your case) to your profiled printer space.<p>

Three other things to consider. First, you're using an incandacent lamp to view your proofs. This light is warmer (more red) than your monitor is likely to be (set to 6500k or D65 so Andrew doesn't get on my case) and that will have an effect. Since you're happy with the colours, that probably isn't bothering you. Perhaps more important is that your monitor radiates light while your print reflects light. The result is that monitor based images have a kind of strength to them that you're not going to find on a print. Lastly, despite what Match 3.2a suggests, even 140 cd/m2 is kind of bright unless your working in bright daylight. In "normal" circumstances, 100 cd/m2 is going to be better. Your monitor profile is based on the brightness so don't go messing with brightness without reprofiling your monitor.<p>

As mentioned above, Adobe's colour engine is better than the Apple one and there is general agreement that you should use that.

Byron, I did read what you wrote and was simply referring to your comment.

 

What I said was that photoshop has a powerful tool that allows you to actually work *as

though you had converted to the

profile* to apply tonal, sharpening and colour corrections necessary for the particular

paper/ink you think you're going to use and see the effects as you work. And you can

go back and forth between your working space and the softproof by just hitting a key

combo. You can also quickly change to different paper profiles to see the effects of using

different papers.

 

Your technique requires you to cancel the convert to profile and then make further

changes to improve the printed file-to-be blindly (without knowing how they'll look) and

then going back to convert to profile to see the effect and then cancelling, make more

blind changes, etc.

 

Do what pleases you but this is not the best advice IMO to give to others. No personal

animosity here, we're trying to help people get the best out of their work methods.

1) Make sure all cartridges are firmly seated in the printer. Do a head check to see if all nozzles are functioning correctly.<p>

2) You have to get your monitor hardware calibrated and profiled otherwise you have no idea of how accurately your monitor is in showing you how the file looks. see

<a href="http://drycreekphoto.com/Learn/monitor_calibration.htm"> here</a> for more info.<p>

3) Your file must be in working space when you open it. Check to see what it is. Adobe RGB is better, sRGB is not acceptable for getting the most out of this printer.<p>

4) Do your tonal and colour adjustments and sharpening (see <a href="http://computer-darkroom.com/tutorials/tutorial_3_1.htm"> this tutorial</a>).<p>

5) Don't follow Byron's advice above about converting to profile. There is a tool in photoshop for doing this which is far more elegant and powerful called soft-proofing. Learn to use it. See <a href="http://computer-darkroom.com/softproof/softproof_1.htm"> this</a> link.<p>

6) In general use Relative Colorometric for rendering intent unless there are lots of saturated colours and when you soft-proof those colours go flat. If so try perceptual as the intent. You can do this right in soft-proof to save paper.<p>

7) Choose Print with Preview in Photoshop and follow

<a href="http://computer-darkroom.com/ps7_print/ps7_print_1.htm"> these </a>instructions.<p>

8) Make certain that the media type you're selecting in the printer driver is correct for the paper you're using.<p>

9) You may need a custom profile made for your printer/papers. Even though Epson profiles are generally good, there is enough pritner to printer variability that the profiles from Epson may not be doing a good job on your specific printer.

According to Bruce Fraser (a recognised authority on colour management), the standards that you're talking about are designed for the graphic arts. I'm not sure about ISO 12646 but ISO 3664 calls for 500 lux for practical appraisal and 2000 lux for critical comparison. If the same holds true for 12646 as 3664 Fraser has this to say:

 

"...the standard wasn't created with monitor to print matching in mind- it mandates that the ambient illumination for color monitors should be less than or equal to 32 lux and must be less than or equal to 64 lux. For monitor to print matching, all these values are way too high- the ISO has acknowledged this and is still working on standards for this kind of match" (pg 210, Real World Color Management, ISBN 0-201-77340-6) Book is highly recommended.

 

Most color management experts that I've read set their LCDs to the following:

1) around 100 cd/m2 depending on ambient light

2) Gamma 2.2 (yes even for apple, the native gamma of the apple lcd range is 2.2 not 1.8)

3) Color temp at 6500k

 

From what I've read, the best technique to compare on screen images to your light box is to try to balance the light box illumination to roughly that of your screen and when looking between the monitor and paper print, either close your eyes for a second or look at some neutral coloured wall.

 

The reason is that a) 5000k on a monitor makes everything look a bit yellow and most people find 6500k a more natural monitor temp and b) your eye/brain quickly adjusts its sense of white balance to compensate for the difference in colour temp dif between the monitor and light box.

<i>Jeez, you'd think Canon would have the camera at least default to 300 or 320 instead of 72.</i>

 

<p>

The camera just records pixel dimensions, eg, 3504x2336 on a 20d. The 320 or 240 or 72 dpi are done by photoshop not by the camera and only relate to printed output resolution. As you up the resolution you decrease the final printed size. You can change the print resolution either in camera raw or in the Image Size dialogue.

I have the 70-200 f4 and just bought the 135 f2 because its black, shorter and thus less conspicuous which makes it a better walk around/travel lens to my way of thinking (I like to be as inconspicuous as I can be). At the moment I either walk around with just the 135, with the 135 and the extender or with a 50 f1.4 also. That gives me essentially a 50-189 f1.4-2.8 which seems to be pretty versatile. For a travel lens presumably you want to travel as light as possible and yet cover a range that you like to work in.

 

The 135 is as sharp as a lens comes though I haven't used it enough to decide if it has the special "glow" that I sometimes I get with the 70-200 (if I was a better photographer I'd have this down but at the moment it seems to just sometimes "happen").

 

One question to ask yurself is how you feel about working with primes vs zooms. I'm discovering that I like how primes make me work to get a shot I like rather than just seeing something, zooming to frame and shooting. I think it makes for more interesting shots but YMMV.

 

The other question to ask yourself is how do you feel about changing lenses or going back and forth with mounting and unmounting the extender. There are no right answers but you have to like the workflow. Considering you'll be traveling and flashing all this nice kit around in various places, in dust, in rain, on busy streets, etc.

 

And of course with the extender the 70-200 becomes a 70-280 compared to the 135/189 combo albeit a rather longer and even more visible chunk of metal at the end of your camera.

Just to finish off and answer your original question, you *can* get better profiles using higher end packages under certain conditions. It's analgous to lenses.

 

Do you get better photos because you buy the most expensive lenses? If you're working at the highest level demanded for professional work the answer is yes but not many people are capable of the skill and knowledge to get those kind of photos. Can you get very good photos with consumer grade lenses? If you know what your doing and you have a good photographic eye, yes but if you compared the subtilties of each print, the professional tools will have some added qualities not possible with consumer grade tools.

 

So in some ways the answer is, how good are you? Do you work with the highest quality lenses, the best cameras, do you professionally develop your images in post processing? If you're working at the highest levels, you probably don't have time to be messing with the intracacies of high end colour management and you should hire a professional. If you profile a lot of paper and you are a dedicated amateur you might find it cost effective to buy a Pulse or eyeone photo. Otherwise you are best off as mentioned above getting a good colorometer and buying some paper profiles from a professional.