Mike Chaney's Tech Corner

In the past, improvements in Profile Prism were usually subtle.

8.1 is not subtle. You will see the improvement immediately, so much so, that I re-profiled the papers I use most.
Like most of us, I settle in a few paper types I like best and we tend to stick with what works.

I use Epson Ultra Prem. Luster and Ultra Prem. Glossy, Epson Ultra Presentation matte, and an HP Advanced Glossy.
I have an Epson R2000 printer, and the results of redoing my profiles with 8.1 are extraordinary.
It's easy too since I saved the target scans and I just load them into Profile Prism, and make a new profile.

If there's anyone who never used Profile Prism and who says, Who needs to make profiles. I use the ones that came with my printer, I am going to attach a screen snap of a scan of a pair of test images I printed. Same sheet of paper; top one with Epson's Prem.Glossy paper and the profile for it from Epson.
The lower image is the same, with my new Profile Prism created printer profile.
By the way, no editing, no tweaking, all default settings.
Both images printed on the same sheet of Epson Ultra Premium Glossy 5*, at the same time, same driver setting,. Only the applied printer profile made by Profile Prism was used for the lower image.

Enjoy!!

Fred

Thanks for the pp update Mike and for the comments Fred.

Regarding the comparison of manufacturer's v pp8.1 profiles. Yes, the difference is striking. On my screen, comparing the bottom left red and green vertical bars of both images, there is a distinct transition "line" on the pp red and to a lesser extent the pp green. No such transition is on the manufacture's image.

Fred, do you have the same type of comparison but pp8 v pp8.1? I would be very interested to see it.

I have reprofiled my 5 Ilford papers and on visually comparing the pp8 v pp8.1 colourspaces there seems to be less at "bright" end, more at the "dark" end and a greater span of the colours.

Regardless of the pp version, I'm surprised at the lack of surface smoothness of the colour spaces of some of the papers. I guess this reflects the very different characteristics of some papers. I don't see this as a problem because i'm very happy with the output of the pp80 profiles.

However I have noticed a possible error with the generation of the 8.1 profile of the smooth gloss paper. The colour space is seriously jagged. See the comparison of 8.0 v 8.1 in the attached image.

I will regenerate the "offending" 8.1 profile as soon as i can.

Regards

Peter

Update: Regenerating the "offending" solved the jaggies problem. The 8.1 profile now looks normal.

 

What viewer are you using?  Attached is my comparison of v8.0 versus v8.1 profile for Epson Ultra Luster on the R1900.  No such jaggies.

Mike

Thanks Mike.

Profile viewed with Mac OS X ColorSync. The jaggies are only on one of the five profiles generated with 8.1

Am regenerating the problem profile now. .....

Update: Regenerating solved the jaggies problem. The 8.1 profile now looks normal.

Peter

Mike.

What Windows profile viewer do you use?

Thanks

Peter

I use PerfX.  I did find a rounding error that could cause some minor jaggies but nothing like what you saw.  I'll release 8.2 that fixes the rounding error just to be thorough.

Thanks,
Mike

Thanks Mike.

I have just compared by printing (via QIU) a single image using v8.0 and v8.1 profiles on two types of paper (with their correct custom profiles).

Unfortunately for each paper type v8.1 is producing less realistic colours compared to: v8.0; the original objects, and; the colour calibrated display.

There is a distinct unrealistic yellow shift in greens and browns. I noticed this in Fred's posted example as well. The colours also tend to be too bright.

Regards

Peter

What kind of test image are you printing?  It's going to be quite dependent on the image, printer, and paper.  The colors in v81 will look brighter on things like gradients and very saturated colors because of the way v81 keeps the colors saturated when dealing with out-of-gamut colors.  v80 was too dull!  But again, I'd like to know what kind of test image.  A lot of people get stuck on mathematical gradients in ProPhoto color space that aren't even possible in real photographs.  For photos, we're finding much better color accuracy with v81!

Mike

The test image is a high resolution still life with grapes, olives, capers, a chilli, and crackers with blue vein cheese and chutney! It's part of a project I'm working on. All natural objects.

If Fred is there: could you please reproduce your comparison with v8.0 versus v8.1?

Thanks.

Peter

Ok Peter I'll try.

The prints actually look far better than the scan of the prints, but best I can do.
The first one is a scan of the test image with PP 800 on top and 801 at the bottom.

Then, I looked hard and wondered if my scanner was brighter at one end than the other.

So I flipped the page and did a second scan with reversed position. Now 800 is the bottom; still the same.
I have to admit that with the actual print in my hand, both look really good, and the distinction is harder to see than what you see on the screen.
Each of us probably focuses on different facets.
Pointedly though, What I especially like in 801, is the improved red. I look at the hood of that car, the guard's uniform, and the British flag, I see a richer red.
Then I look at the spools of thread. 801 is showing me a better distinction between the red and the Orange next to it.
I look at the yellow flower (sunflower, I guess).   I see a refined yellow with petal detail in 801.
800 is a tad brighter yellow and hides the detail.

There is also a bit more shadow detail in the brownish cabinet (I guess it is a cabinet) in 801

I will be home after lunch for the rest of the day. If you would like to pop in, I buy supper.

Fred

PS This is Epson Ultra Premium Luster paper

Thank you very much Fred.

I appreciate the time you put into the comparison. It is interesting.

I understand what you are saying.

What I am looking for in a profile is one which results in a print which most closely represents the actual colours of the objects photographed. This is especially important for fashion garment photography. Any colour enhancements are due to my editiing and not automatically applied by the profile.

In my test image (described above) all objects look natural (as I intended) with the 8.0 profile but with 8.1, capers look too yellow, crackers unnaturally yellow and the red jalopeno not "real jalopeno red" anymore but too bright a red. Grapes are also to bright.

In 8.1 objects may look more eye popping but i want to determine when something is to be eye popping.

If I have time I will scan the test prints and post them.

The test image I have been using in the print tests is actually the "Late Harvest" image on this page: http://www.pastoralsystems.co.nz/photography/galleries/raw-deal/ (http://www.pastoralsystems.co.nz/photography/galleries/raw-deal/).

.... the test comparison I have been referring to is now attached. The reduced jpg doesn't do the image justice. Top is v8.0 and the bottom v8.1.

Regards

Peter
 
      

Hi Fred & Peter,
I am not finding that to be the case.
I made new profiles with 8.1 for both Iford Smooth Gloss and Epson Archival Matte. I made prints with the test image that Fred used and the "Printer Test File" that come with PP - it has a hand with coloured balls, a lady in a pink top and a B&W dog with a monitor.
I made comparisons between PP 8.0 and 8.1 for the "Printer Test File" where 8.1 showed reds that tended towards orange on Ilford smooth gloss. Flesh tones looked like they had a yellow tinge. The PP8.0 print was weak on the reds and still tending towards orange with reds but much less saturated.
On Archival matte I compared PP8.1 to the Epson profile. The latter is not perfect and I would prefer more saturation but when compared to PP 8.1, I would make the same comments about flesh tones and reds as above; this was the case with the "Printer Test File" and the one Fred used (baby face and cotton reels etc.)
Finally, when the PP8.1 prints are compared to the (calibrated) monitor it confirms my comments above, ie. reds tending to orange, flesh tones with a yellow tinge, not sure about grass colour either, towards yellow a little.
I did make a scan of 2 prints but it did not show the differences properly so have not posted it.
Terry

I also appreciate your thoughts too. That's how Mike perfects his stuff.
There's another point to consider.
This is bandied about a lot. My prints do not match the screen; usually followed by the prints are too dark!

99% of the time, the monitor is too bright as you know!
My point is that I hope my prints match my screen and so do you hope for the same.
Yes, our monitors are profiled, but us the contrast and brightness the same for each of us?
Here's the tricky part.
If I use Mike's printer profile, by prints are so close to what the screen shows, that I can give it an A. (I agree that 801 is pushing the envelope on saturation)
If I use the Epson Ultra Prem Glossy profile, my screen and the print are not even close.
If I use the Epson Luster profile, pretty close.

Next point of discussion. Reality! 
I can shoot a flower and look at the print and I get a match... if I walk outside to compare my print to the flower and it is tomorrow, and the color changed...
So we tend to use test images with a steady unchanging vast combinations of colors and we can only compare an 800 print to an 801 print.
I try not to fixate on one color.

It's fun to share.
Last comment. I made my 801 profile with no tweaks at all. I will however, most likely back off to a minus 2 saturation for my "keeper" profile.
That is unless Mike performs his magic.

Fred

Terry,
As you well know, some papers are not easily profiled. Archival Matte is one and I have only tangled with Ilford Smooth Gloss in the past....
OK you point is taken, but when we encounter more difficult profiling, we have the ability to EDIT and adjust the profile so the paper performs well.

Fred

Thanks for your comment Terry M.

Your yellow tinge comments are consistent with my observations, also the red issues.

I have delved into accurate colour matching of prints to actual objects to a considerable extent when I have done fashion garment photography where fabric colour is critical. To this end PP8.0 has served me very well, but at this point v8.1 doesn't seem it will.

We will see what Mike comes up with.

Well it's now midnight for me!

Regards

Peter

Gents, just got 8.1 to use with my r1900. What are the correct printer driver setting s to use to print the target image?

Peter,

On the left is the exact image set you posted here: top/bottom 8.0/8.1.  I took the liberty of copying the image off your website and set that to the right.  Are you telling me you think the top (8.0) image on the left is closer to the one on your website than the one on the bottom (8.1)?  If so, I either need eye surgery or brain surgery.  Please tell me I shouldn't book an appointment with one of those two surgeons!

What am I missing?  The top one looks dead: the grapes almost look like coffee beans.  They're almost all one color.  On the bottom, the pepper is much closer to what is on your website (which is pictured by itself on the right) in every region: grape have actual separation in color and they are more "grape" color and less "cocoa".  The crackers and the spread are almost dead on with the bottom left (which you say is 8.1).

Don't really know what to say with this comparison???

Is the image on the website "baked" for extra saturation and the one you printed not?  If so, I'd like to see a small version of the original here so we have something to compare to.

Also, you should probably try 8.3 to see if you see any difference.  I made some small enhancements to the saturation algorithm that could affect some printer profiles, particularly if you are using a scanner with lighting (like CCFL) that can cause metamerism or one that has a particularly small gamut both of which can affect the scan of the targets.

Mike

Hello Mike.

No you don't need to book an eye appointment. Relax!

"Are you telling me you think the top (8.0) image on the left is closer to the one on your website than the one on the bottom (8.1)? " ..... no I am most certainly not. I merely linked to the web version to show the image I was verbally describing.

You are comparing a scan of printed images with a jpeg web version, not printed nor scanned. The scanned version is not as bright as the original print (I've never calibrated the scanner), actually quite dull. A comparison is not relevant.

If you saw the printed large version using v8.0 you would instantly see what I mean and accept that the v8.0 print is much more realistic cf 8.1.

All my comments have been been directed towards the quality of the end print.

When I took the photograph I printed it and compared it to the actual objects and as viewed onscreen with the editor. I was (and still am) very happy with the v8.0 print as it was very close to the actual objects under the lighting I used, and what I saw onscreen. Hence I consider the v8.0 print as a good benchmark. Prints using v8.1, and to a lesser extent v8.3, have deviated from the realistic colours. The yellowish tint was the most glaring problem with v8.1.

The "realism" check I described above was also applied to the other three images of the series that are shown via the link. Each was taken at different times with non-identical lighting.

Hope this helps.

Regards

Peter

PS The scanner I use is an Epson Perfection 4990 Photo.
  

Hello Doug.

The online Profile Prism help will tell you all you need to know.

Good luck.

Peter

Doug,
In 99% of making printer profiles, the driver is set to BEST Quality, select the correct matching paper, and then set the driver to NO COLOR ADJUSTMENT  (OFF)

It is important to remember what settings you use to create the profile, should be the same settings when you use the profile.
Fred

Hello Fred.

Sorry to bother you again, but I find this whole topic interesting and intriguing.

Would you be willing to generate and present a v8.0 versus v8.3 comparison of your test photo?

Thanks

Peter

Of course.
I am not a fan of scanning a pair of prints when the object is so subtle as to be lost by scanning and then resizing that to fit within the post guidelines.
I would be happy to send you the test images via email and you can make your own A/B comparisons.

Just an aside:  Mike's comments about perhaps needing glasses after looking at your comparative prints was exactly my initial reaction when I read your post and your choice of which looks right!
I just didn't have the courage to say it.

Let me have your email address, please.
If you don't want it public, then email it to me  wathree.ssz@verizon.net

Thanks,
Fred

Fred, Thanks. My question is; I selected no color adjustment but then it allows the selection of Adobe RGB or Epson standard or vivid color space and further selection of gamma. i set it to Epsosn standard and gamma 1.8 (per what I read in the help). Are these correct?

Ahh, but now I want you to think about something.  Something regarding how very important and relevant that scan really is!  The scanner is the same scanner that scanned both targets when you created your printer profile.  That means as the scanner sees it, v8.1 did a better job matching the image that is on the website... and that's the same scanner that "saw" those two targets you scanned for your profile.

In other words, in the eye of the scanner and under the scanner's light source, v8.1 is more accurate.  Now it'd be interesting to see an IT8 target scanned in the same scan as both prints so the scanner could be calibrated but Epson Perfection Scanners aren't known for drastic color/saturation shifts so I'm assuming your scanner is "normal" and produces reasonably accurate color.

What I'm getting to with all this is that your scanner uses CFFL lighting (fluorescent lighting).  That light source causes very significant metamerism on the IT8 target which shifts colors all around on the IT8 target and in turn affects the printer profile.  Unfortunately IT8 targets were meant to be shot under D50 lighting or at least full spectrum lighting and those IT8's are subject to metamerism.  We switched to Canon LiDE scanners almost a decade ago because their LED light sources are much more "neutral" from a full spectrum standpoint.

Now, I'm not suggesting you go out and buy a $79 LiDE scanner but I am well aware of the downsides of creating printer profiles with cold cathode scanners.  They almost always require some adjustment after the fact because no two CCFL tubes are the same.  It's possible the old v8.0 had some idiosyncrasies in its color matching algorithm that favored certain scanners but I can guarantee that v8.x is more accurate in general (that is, for the average user).

Also keep in mind that many, many colors in your image may not be reproducible on your printer.  You'd be surprised how many colors you see on your screen simply cannot be printed on your printer.  Printers have a much narrower gamut and dynamic range and while your printer may be able to produce a certain hue and saturation (like a deep red), it may not be able to reproduce the brightness of that color at the same time because inks on white paper can only be/go so dark or bright.  So compromises are always being made in how to represent a color that cannot be reproduced on your printer.  You can lock hue and then get a compromise of saturation and brightness.  You can lock the saturation and then fiddle with brightness and hue a bit.  Etc.

I may add something like a "vibrancy" checkbox in the next version where, if you UNcheck that box, it doesn't favor saturation as much.  That would tone down the saturated look at the expense of a small amount of hue shift (more like the old 8.0).  But for now, you should be able to just choose something like a -3 on the saturation control on the main window before developing your profile and see how that looks!  That should get pretty close to the old 8.0 and still have more accurate hue.

Regards,
Mike

No Doug!
You need to check NO COLOR ADJUSTMENT.
I have an 1800 driver in front of me here, and you need to click on ICM, and then wait. The NO COLOR ADJUSTMENT box will appear. Check that!
After you check that box, all those options disappear. 
Think this way.    You are trying to tell the driver: Hey! I am turning you off and I do not want anything changed from what Qimage (or any other software) is sending to be printed.

Fred

OK, I got confused by the help file as it discussed setting gamma etc. I will follow as you suggested.

Hello Fred.

I understand your avoidance of the tedium of scanning, converting and publishing.

The reason why I asked if you would generate the images was that it standardises the process so a relative comparison can be made across all the test images you have produced. An v8.3 scan is the only one missing from the set. 

Regarding your first comparison of Epson versus ppv8.1 profiles. I question whether Epson would produce a profile that results in a dull rendition of such a test image. Unless your printer has deviated from spec, i assumed that the extent of the dullness of the Epson profile scan was the result of scanning and conversion/compression to the JPG you posted. If this is the case then this would have affected both your Epson and v8.1 posted images.

This is certainly the situation for my 8.0 versus 8.1 scanned comparison. The scanned to JPG images are dull compared the actual prints. Hence what I consider the better rendition, the v8.0 print, is seen onscreen as dull whereas what I consider slightly over bright (ignoring the red and yellow issues reported by me and Terry), the v8.1 print, is seen as more "acceptable". My v8.0 print is actually closer to the image posted on my web site.

Hence I conclude that with the scanned and posted test/sample images the relativity between tests can be best judged and to a lesser extent their absolutes.

The smoothness of the transition from bright to dark of the various colours in your test image are useful. If the transition is smooth in the original test image then, for my use, it should be as smooth as possible in the printed image. This doesn't seem to be the case with 8.1.

For my type of photography if the subject is bright and vibrant then i expect the printed image to be bright and vibrant. If the subject is dull then I expect the printed image to be dull. If I want to deviate from those "default" outputs then I make adjustments in the editing software. I guess what I am saying is that i want the default processing steps in my workflow to be as neutral as practicable.

For years QImage and Profile Prism have been critical steps in my workflow to acheive the above.

Regards

Peter

Thanks for your full and detailed reply Mike.

I understand your explanation. It was good to revise those points in such a concise form. I have always realised how important the scanner is and the technical compromises involved especially relating to colour.

Currently my main use of the scanner is for documents and for colour profile generation using PP. If I were to regularly scan photographs and was wanting very accurate colour rendition then I would calibrate the scanner. In the meantime I have assumed the default quality of the scanner to be high enough.
 
I have always believed that PP uses the IT8 target as colour references to generate the profile and, in the process, to correct as much as possible the idiosyncracies of the particular scanner, including the effects of the CFFL. Is this correct?

I believe that my Epson 4990 has performed its step in PP profile generating process quite well (and so it should). During processing the IT8 and Target profiles have always been close to the ideal profiles shown in your PP help document. Furthermore the brightness ranges have been "full" (with very few exceptions) and the number of excluded cells minimal.

Clearly there are different expectations of printed output: those who want automatic "pop" and those who want "neutrality" by default. For that reason I think that your suggestion of the option button the allows user to choose the "built in vibrancy", or not, is a good one.

To my eye, the PPv8.3 printed output looks better than v8.1.  

Regards

Peter

Peter,

Your feedback is much appreciated.  I've just released my final adjustment in v8.4.  With v8.4, Profile Prism will generate profiles as "mathematically" accurate as possible.  If you want extra "pop", you can add it easily by selecting +1, +2, etc. in the "Saturation" control on the main window.  I've run v8.4 through a lot of paper and with a good scanner you should see stunning color accuracy!

Tip: Perceptual rendering intent tends to desaturate colors a bit in order to expand the usable gamut of the profile beyond that which is reproducible by your printer.  So if you want the highest color accuracy possible (within the limits of your printer/paper/ink), re-generate your profiles using v8.4 and print your photos using Relative Colorimetric intent.

Mike

Thank you Mike.

I am looking forward to testing v8.4. It sounds great!

How does Relative Colorimetric intent in the profile interact with, or relate to, RC set while printing with QIU? I always print with RC set.

Since I'm in early Saturday afternoon, I'd better get outside and enjoy the day. Will install and test v8.4 tonight.

Regards

Peter

When you create a profile in Profile Prism, it automatically creates two profile look-up-tables inside the profile (ICM file): one for Perceptual and one for Relative Colorimetric.  If you choose Perceptual, it uses the perceptual table in the profile.  If you use RC, it uses the RC table.  They are simply two different mappings that map the colors completely differently for different purposes.  It's sort of like having two different profiles in one ICM.  Actually, exactly like having two versions of the profile.

If you already use RC, you're good to go: that's typically the recommended setting for printer profiles.  Only reason I pointed it out is if you happen to use Perceptual, (some) colors might be a bit muted as the profile tries to expand the gamut coverage of your printer.  Perceptual intent does that by toning down saturation a bit (in needed areas) to "fake" the appearance of the printer being able to resolve colors beyond what it really can.  Why would you even want to do that?  Only in special cases where you might want to show a color gradient in a logo or something that might have very bright out-of-gamut colors and it's more important that the gradient show than match the exact color.  In a case like that, RC might clip the colors and give a rough gradient.  Perceptual would tone them down a bit (typically desaturate) so that you at least get a smooth gradient as opposed to one that might look blown out.

Oh and one other "nasty" little tidbit about ICC profiles is that they are not "smart".  They can't read the gamut of the input image and only scale (fade) colors for perceptual intent for only the colors in the original image that need it.  ICC profiling is done with no consideration of image content... which means with perceptual intent, the gamuts are "smashed" the same amount for every image, even if you only print an image that has very subtle colors: those will look faded too with perceptual intent.

It's 10pm here so I'm done for the night.  Been at it 14 hours today.  ;)

Mike

Peter,
Your points are so well taken that you have to appreciate the stark wonder I experienced when making the initial tests.
I felt the same way.
Could this be??????????

I would call your attention to this.

http://www.ddisoftware.com/prism/about.htm

Mike made this one, not me.... I say that with the knowledge that Mike wouldn't trip himself up with some error that I might make.
See my screen snap of his scanned comparison.
SAME BASIC DIFFERENCE between the Epson Profile and the Profile Prism. Mike uses the term Manufacturer's profile because he found there are others that have poor profiles too.
I was dealing with Epson Ultra Premium Glossy, supposedly the top Glossy paper on the planet.
I was astonished!!
Then Mike switched to Epson Ultra Premium Luster (My all time favorite paper) so I switched too, and the comparison is still there, still the same problem, albeit, not quite as drastic as the Glossy. See screen snap attached, and then go to  http://www.ddisoftware.com/prism/about.htm

I apologize for not supplying the scans you wanted.... I have a desk full of test prints, scanned prints, A/B prints, loads of Beta versions prints, which I can barely see over the top.
I was at it too for 12 hours yesterday  ;D ;D ;D

If you still want any scans, I will try to find them in the pile. 
I think the final 8.4 which has been tweaked back, plus the ability to use Relative Colorimetric to put back the color that was desaturated (washed out) because of the way Perceptual works, you have the best of both worlds.
I think this 8.4 puts all the 8.1 and 8.2s to bed.
When I was testing yesterday, and we got to the RC setting, I remembered that I used to use a lot of Ilford paper. They supplied profiles, and their settings for Glossy,  included Relative Colorimetric instead of the common setting pf perceptual.

Glad to get together.

Fred

Thanks very much Fred.

In Mike's comparison the manufacturers profile is totally believable!

As you will be aware, Mike described Perceptual v Relative Colourmetric intents before he went to bed! I'm familiar with them and I always use RC when printing.

Maybe I have misunderstood some of the comments relating to intents. Are you actually selecting RC somewhere in Profile Prism when you are setting up and generating a profile? If so, where are selecting it?

I'm working on the 8.4 profiles now so if you can reply soonest that would be appreciated.

Thanks again.

Peter
 

Hi Peter,
The place to select your preference of Intent, is in Qimage
Profile Prism creates the profile with all built in.
See attached
Notice that the RC is showing in the second screen snap indicating I selected Relative Colorimetric

Fred
Just to clear confusion.
Mike and I live in the same Geographic time zone, but I am three hours earlier than Mike's awake zone. 8)

Thanks Fred.

That's what i thought but some of the comments made me think otherwise and that i was missing something relating to PP.

Yes, I'm familiar with the Intent settings in QIU. The settings in your screen captures are as I've been using for "yonks".

Now to print my test image with 8.4 .....

Regards

Peter

Good to talk about it as there are many reading these posts that are saying, " Change to RC? WTH? Where is that?"
They might not have seen that before.

Fred