Rendering intent-best print preview?

JC

In Gimp and Scribus, I have two rendering intents that are considered to be the best for display of photos:

Perceptual
or
Relative colorimetric

I make a calendar every year, and we sell about 500 copies. We currently do the printing with Uprinting because we've gotten the best price from them. I do any photo editing necessary in GIMP and do the layout in Scribus. The customer service reps at Uprinting I've talked to have never given me any specific printer profiles to use for softproofing (they use 24" Komori or 40" Mitsubishi offset presses), so I just use Scribus's default for CMYK coated. I have Scribus convert the RGB images to CMYK and export the files as PDFs for sending to the printer. The printer does no color correction., and we get no hard copy proof.

Most of the pictures usually turn out ok, but last year two of the images lost a lot of color vibrancy, and one lost a lot of detail in subtle shadings of pastel orange.

Uprinting does large volume commercial prints, so I can't just order a handful of the photos printed as proofs.

Ok, finally to my question, when viewing the photos on my screen, should I be using perceptual, or relative colorimeteric? I have only been able to find info that says, essentially, 'in some cases perceptual is better, in some cases relative colorimetric is better', but never really specifies parameters. For most photos it doesn't make real difference, but for some it does. I have a dell IPS monitor that is calibrated well for color, but not so well for brightness.

Thanks!

arodney

kolibri wrote: »

In Gimp and Scribus, I have two rendering intents that are considered to be the best for display of photos:

Perceptual
or
Relative colorimetric
!

Which ever looks better in the soft proof. It is image specific and when woking with a Perceptual intent, the qualities of the ICC profile (all perceptual renderings are vendor specific). ICC profiles don't know squat about images or color in context. You have to look and decide.

Sam

Kolibri,

Perceptual, and relative colorimetric are two common rendering methods.

Relative colorimetric only deals with those colors that are out of gamut. It only changes / modifies those colors that are out of gamut and leaves all the other colors as is and can result in some color clipping or strange transitions.

Perceptual will not only modify those colors that are out of gamut but will modify other colors in an attempt to produce a smoother transition between the out of gamut and the in gamut colors, sometimes resulting in a less saturated image.

If all colors are within gamut you shouldn't see any difference between the two.

I called Uprinting and their calendar prices are really good for 500 to 1000 pieces.

As for ordering you can submit .jpg files in srgb and let them do the conversion to CYMK. They should be able to do a better conversion for their printers than you can.

Second you can get a hard proof of the printed images (you don't need the month text page) for a cost of $5.00 per page. That's $70.00 for all images including front and back. A wise choice. This only increases your cost by $0.14 per calendar to insure the final printed calendar displays your images as you intended. I would never print a large quantity of anything without a hard copy proof.

Sam

arodney

Sam wrote: »

As for ordering you can submit .jpg files in srgb and let them do the conversion to CYMK. They should be able to do a better conversion for their printers than you can.

Actually no. Not if it is the same profile provided to you that they actually use. Which is what it should be. You would then be able to pick each image's rendering intent based on a good soft proof (assuming the display is properly calibrated and profiled). And if they don't use Black Point Compensation which is possible depending on how the conversions are made, it is possible a conversion using BPC would be significantly better.

If they and I use the same profile, CMM and settings, we get identical results. Theirs is no better.

Sam

arodney wrote: »

Actually no. Not if it is the same profile provided to you that they actually use. Which is what it should be. You would then be able to pick each image's rendering intent based on a good soft proof (assuming the display is properly calibrated and profiled). And if they don't use Black Point Compensation which is possible depending on how the conversions are made, it is possible a conversion using BPC would be significantly better.

If they and I use the same profile, CMM and settings, we get identical results. Theirs is no better.

In my conversation with them they have no profile to provide. Hence you can not soft proof the image. I don't know of any way to get a good match without a specific icc profile for the paper and printer.

I have no idea if they are using BPC or what rendering intent or CMM they are using. Maybe my CMYK conversion would be as good as theirs, but it's their printers, they should have a better conversion.

Since I can't figure out EXACTLY how they are dealing with the image files, but the OP said they were pleased with all but two images it is my feeling that a hard proof will be the ultimate example of what the calendar will look like.

Sam

arodney

Sam wrote: »

In my conversation with them they have no profile to provide. Hence you can not soft proof the image.

Nor convert to CMYK. So the question of who'd do a better job is moot if they can't supply a profile. Some of us could build one but it would be quite costly to do so.

Sam

arodney wrote: »

Actually no. Not if it is the same profile provided to you that they actually use. Which is what it should be. You would then be able to pick each image's rendering intent based on a good soft proof (assuming the display is properly calibrated and profiled). And if they don't use Black Point Compensation which is possible depending on how the conversions are made, it is possible a conversion using BPC would be significantly better.

If they and I use the same profile, CMM and settings, we get identical results. Theirs is no better.

arodney wrote: »

Nor convert to CMYK. So the question of who'd do a better job is moot if they can't supply a profile. Some of us could build one but it would be quite costly to do so.

Andrew,

Just trying to provide the best way to get what the OP wants. The price per calendar is only $3.00 to $4.00 each in larger quantities.

One offs from a photo lab will run you $12.00 to $16.00 or so.

I little wrangling to get the image the way you want could easily be worth it.

Sam

JC

Sam wrote: »

Kolibri,

As for ordering you can submit .jpg files in srgb and let them do the conversion to CYMK. They should be able to do a better conversion for their printers than you can.

Second you can get a hard proof of the printed images (you don't need the month text page) for a cost of $5.00 per page. That's $70.00 for all images including front and back. A wise choice. This only increases your cost by $0.14 per calendar to insure the final printed calendar displays your images as you intended. I would never print a large quantity of anything without a hard copy proof.

Sam

thanks Sam, and arodney. I didn't mean to post and run, but I got called out of town for work.

I usually submit pdfs, we have a fair amount of text and some vector art on the calendar, and I think that pdfs would render better.

Lets see if i can summerize the thread,

Since they have no profile to give me, I should keep the images as RGB when I convert to pdf, and adjust the colors, saturation and luminosity/intensity to look good on my semi-calibrated monitor, and just do a bit of trial and error with print proofing.

Would printers from a local print shop, like kinko's be an approximation for what their printers produce?

Sam

kolibri wrote: »

thanks Sam, and arodney. I didn't mean to post and run, but I got called out of town for work.

I usually submit pdfs, we have a fair amount of text and some vector art on the calendar, and I think that pdfs would render better.

Lets see if i can summerize the thread,

Since they have no profile to give me, I should keep the images as RGB when I convert to pdf, and adjust the colors, saturation and luminosity/intensity to look good on my semi-calibrated monitor, and just do a bit of trial and error with print proofing.

Would printers from a local print shop, like kinko's be an approximation for what their printers produce?

Arodney is the real expert here, but that won't stop me from chiming in.

With the information you provided I can't see any scientific way to even start with.

There is no such thing as a semi-calibrated monitor. It does sound like you are squeezing the colors down to fit into a CYMK output gamut, so maybe you could try to convert to CYMK and see what the results are.

Since you are starting with an unknown (monitor) you can't be sure that what you see as a processed image is really what the numbers represent you can't know what your CYMK conversion will look like printed, but you should be able to see a shift between your rgb monitor image and the converted CYMK image. The larger the shift the more you need to worry.

By sending in the images (in whatever color space) and getting a hard proof you will see what the actual / real calendar will look like and if you have say 8 images that are acceptable, you can take a close look at the remaining 4 and see if you can tweak them or replace them with different images.

Yes, this will take some time and may need to be repeated several times, hopefully with fewer and fewer retests.

Sam

JC

Sam wrote: »

There is no such thing as a semi-calibrated monitor. ,,,

Since you are starting with an unknown (monitor) you can't be sure that what you see as a processed image is really what the numbers represent you can't know what your CYMK conversion will look like printed, but you should be able to see a shift between your rgb monitor image and the converted CYMK image. The larger the shift the more you need to worry.


Sam

My monitor is calibrated well for color/hue, not so much for intensity. I have 1 good IPS monitor a crappy widescreen at home and two laptop monitors. I have them all calibrated with a lower end Spyder, and they are consistent in color and hue, but not in intensity. In some monitors I can see detail in areas that in other monitors are deep black and washed out white.

The orange popsicile colors of canyonlands are the only time I've been unhappy with the colors, my main issue is witht the loss of intensity in some of the images. I know that printers and screens have different dynamic range.

I did all the proofing on my IPS monitor, but the prints all seemed to have less dynamic range, which only really mattered on the images with larger areas of dark color, which came out much flatter, although relatively true to hue. Does that make sense?

Sam

kolibri wrote: »

My monitor is calibrated well for color/hue, not so much for intensity. I have 1 good IPS monitor a crappy widescreen at home and two laptop monitors. I have them all calibrated with a lower end Spyder, and they are consistent in color and hue, but not in intensity. In some monitors I can see detail in areas that in other monitors are deep black and washed out white.

The orange popsicile colors of canyonlands are the only time I've been unhappy with the colors, my main issue is witht the loss of intensity in some of the images. I know that printers and screens have different dynamic range.

I did all the proofing on my IPS monitor, but the prints all seemed to have less dynamic range, which only really mattered on the images with larger areas of dark color, which came out much flatter, although relatively true to hue. Does that make sense?

What brand model of IPS monitor do you have? I would only use the IPS monitor for image processing and printing.

If your IPS monitor is displaying washed out (blown out whites, not showing to be blown out on the histogram) and blocked up blacks you will not be able to get a precise print.

Also luminance as well as contrast is important to get right. While there may be no real absolute with regard to luminance and contrast you can bet the factory settings are too high. luminance values for printing will normally fall somewhere between 80 cd/m2 to about 115 cd/m2. Contrast will normally be say 200/1 to 300/1. White point is important as well and most seem to calibrate their monitors somewhere between D50 and D60.

That said, with what you have told me I would need to take an empirical approach. Do the best I can with the images and send them in for a hard copy proof. Compare the hard copy proof with the monitor and adjust the image as necessary to get the results I am happy with.
Again CYMK is a fairly small color space and can not reproduce many colors you can see on your monitor. So you may want to consider a different image if you can't get the CYMK conversion color to your liking.

Sam

arodney

Rendering intents are image specific! Each profile package will produce often a striking difference from another (from the same device) when using a Perceptual rendering as this is vendor specific. You have to view each image and toggle between the two and select the one you prefer. Or just pick one blindly. But profiles don't know squat about images or color in context. Everything to them is a solid color that eventually has some Lab triplet value or similar value the profile then converts to what it 'thinks' should be the new color value for the output device and in that device color space.

JC

arodney wrote: »

Rendering intents are image specific! Each profile package will produce often a striking difference from another (from the same device) when using a Perceptual rendering as this is vendor specific. You have to view each image and toggle between the two and select the one you prefer. Or just pick one blindly. But profiles don't know squat about images or color in context. Everything to them is a solid color that eventually has some Lab triplet value or similar value the profile then converts to what it 'thinks' should be the new color value for the output device and in that device color space.

Thanks Rodney. By profile here, I'm assuming you mean the profile a program uses to convert between RGB and CYMK?

I"m trying to get this whole process set up with templates so that whoever inherits this extracurricular project can just plug in the images and new text, convert to PDF, and send to printer if they don't have any interest in delving into the arcane arts of image post processing.

I"m using scribus, and I think exporting as a jpeg opens up too many potential issue, so I think I need to stick with PDF.

I don't get to pick and choose the images, we have an internal contest and the top 13 get used. This year we have a couple of great ones, some good ones of great locations, some HDR, oh, and mine. I have to make them work, and next year this has to be an easy process.

My monitor is a Dell IPS, lower end model, it doesn't list it on the front, probably the back, but I can't get to it now.

So, stick with relative colorimetric for the templates?

arodney

IF you are forced to make blind color space conversions (this could be RGB to RGB or RGB to CMYK), then all you can do is select a rendering intent and move on. If I had to pick one, I'd probably stick with Relative Colorimetric. I say this based on using a lot of different profiles from different packages, soft proofing and viewing the two options. Most of the time, I visually prefer the RelCol intent so that's what I'd pick if I couldn't decide on an image per image basis.