Imageprint 10 and the Canon Pro 1000 - not getting the advertised result

[traderjay]

I downloaded the demo version of the Imageprint 10 because according to the reviews, this RIP will produce better results than the standard drivers. I had high hopes for this software because my printer has been busy everyday since the birth of my newborn as it is our tradition to document his first 100 days.

So naturally the photos consists primarily of a newborn closeup shots under natural daylight and is very demanding on the printer due to the almost translucent like skin tones. My camera is the 5D Mark IV using either the  Canon 24-70 F4 L or the 70-200 F2.8 II IS L lens with ISO ranging from 100 to 800 depending on the available light.

The paper that I am using is either Ilford Smooth Gloss or the Smooth Pearl. My workflow consists of DxO Optics Pro to Print using manufacturer provided ICC profiles, and a custom calibration using the printer's built in calibration feature. I also use Canon's Print Studio Pro via Photoshop CC.

On the Imageprint side, I selected their custom profile based on the two types of paper that I have. I did about 5 test prints of various photos, each with different types shadow, bokeh and skin tones and placed the images from Imageprint and Canon Drivers side by side on a evenly illuminated desk with a high power magnifying glass. Initially I can barely see any differences between the two and certainly not the improvement as advertised on their website or by an online review ronmartblog. In certain instances, prints done by imageprint shows faint but visible banding in skin tones that are naturally lit by daylight. I looked even harder in the shadow areas and cannot see any difference whatsoever. There are also virtually no color differences between the two either.

I know that Canon provides 16 bit drivers out of the box so maybe it has already been optimized to the printer's best ability and there is really very little you can squeeze out of it?

Below is the image showing the worst banding:

[Sbarroso]

I have no experience at all with imageprint nor DxO software. That said:

The first question I have is,  are you satisfied with the outcome of canon's driver (+ canned profile)?
If yes... Then, don't expect much more.

The second is: is the banding to something aligned with the head/ rollers? Or is banding due to a veru slight change of tone? The later would be a sign of 8 bits output. Canon print studio pro should be able to work in 16 bits (using the xps driver in the case of my printer). Note that nor Lightroom nor Photoshop can print 16 bits directly, even if using the XPS driver. In my case that's running Windows and pixma pro 1. I must go first to plug-in (but I do it only if 16 bits is really necessary).

Regards

[traderjay]

I have no experience at all with imageprint nor DxO software. That said:

The first question I have is,  are you satisfied with the outcome of canon's driver (+ canned profile)?
If yes... Then, don't expect much more.

The second is: is the banding to something aligned with the head/ rollers? Or is banding due to a veru slight change of tone? The later would be a sign of 8 bits output. Canon print studio pro should be able to work in 16 bits (using the xps driver in the case of my printer). Note that nor Lightroom nor Photoshop can print 16 bits directly, even if using the XPS driver. In my case that's running Windows and pixma pro 1. I must go first to plug-in (but I do it only if 16 bits is really necessary).

Regards

I am happy with the Canon driver + DxO optics pro or Canon Print Studio pro workflow but I thought I can take it one level further with the famed Imageprint. I also suspect that working in JPEG could be the factor as well since IP cannot work with RAW files directly.

As for the 16 bit print driver support - i cannot see a difference in quality from prints made in Canon Print Studio Pro or DxO, which prints via the XPS driver.

[Ryan Mack]

JPEGs are 8 bit which is probably the cause of the banding you are seeing. I would recommend outputting 16 bit TIFF files with no alpha channel in the AdobeRGB or ProPhotoRGB color space for printing in ImagePrint.

[dchew]

Yeah, when I output to ImagePrint (IP) I create a sized 16 bit tiff in prophoto.

I say sized because of output sharpening. IP10 has sharpening but I just don't know it as well, so I use what is familiar to me in LR and/or PS.

Worth a try outputting tiffs while you have the demo. I would be surprised if you see a big difference using IP vs traditional editing software + print driver. Of course, "big" is subjective. I use IP more for the workflow than the improved print quality. On my printer (Epson 7900), there are some b&w permanence improvements using IP.

Dave

[traderjay]

I tried with this test image - http://www.digitaldog.net/files/Gamut_Test_File_Flat.tif

and it seems IP is not recognizing or importing the file properly. In the 14 ball test the black transition is showing black circles instead of gradual transitions...

[dchew]

I tried with this test image - http://www.digitaldog.net/files/Gamut_Test_File_Flat.tif

and it seems IP is not recognizing or importing the file properly. In the 14 ball test the black transition is showing black circles instead of gradual transitions...

Ah, "Bill's Balls." I did that test when I first started using IP 9. Bill's balls looked the worst vs. LR+my custom profile (DChew), and maybe worse than the canned Epson profile for Canson IBP (CIBP). I still have those three test prints:
Balls from L to R, ranked best to worst:
Black/White: IP>DChew>CIBP
Cyan: Tie, with CIBP a bit better; none look very good.
Magenta: CIBP>DChew>IP
Yellow: CIBP>DChew>IP; IP looks horrible on the bottom ball which has stepped rings.
Red: DChew>CIBP>IP
Green: DChew>IP>CIBP
Blue: DChew>CIBP>IP; IP looks horrible, especially the top ball which has stepped rings. Bottom ball is almost black.

I posted that here a few years ago, and Andrew chuckled. Keep in mind Bill's balls are specifically set up to test the edge of the gamut and how colors are mapped at those edges. In my (very) uneducated opinion, I think these tests are analogous to lens tests. Lots of different ways to perform the tests to highlight various attributes, and the proof is in the printing, not one specific test.

After doing that test, I concluded IP is very good for b&w printing, and for a stock of reasonable profiles if you use a lot of different papers. If you are printing color on primarily one or two paper types, then I don't see it making much sense; just build your own profiles for those few papers. Again, that is results-only benefits. I do see other benefits with IP, but those are very specific to my workflow and personal strengths/limitations.

Dave

Edit: Here is that original thread: Print quality shootout

And a follow up last year: ImagePrint Review and Demo

[traderjay]

Thanks for the reply dchew and maybe I can come to the conclusion that IP10 is completely unsuitable for Canon users:



I have a crummy scanner but you can still still the day and night difference between the two. On the actual canon print, the black to color transitions are completely smooth without any banding. It makes me question the validity of this review - http://www.ronmartblog.com/2017/01/review-colorbyte-imageprint-10-discount.html

On the other hand, maybe Epson's stock drivers are subpar and isn't coded to fully unleash the printer's potential, hence the majority of the users that see a big difference are Epson users.

[dchew]

You may be right, but that is a big conclusion to draw from one or two paper types in color. BTW, did you ever print those images you were struggling with as tiff outputs to see if the banding went away? I only ask because like your results with Bill's test balls, I was kinda shocked at first. As Schewe pointed out in that first thread I linked to, this type of test may not be all that relevant. Over the past two years, with several direct comparisons, I do not detect problems with color prints through IP. Of course, I may not know what to look for.

On the other hand, maybe Epson's stock drivers are subpar and isn't coded to fully unleash the printer's potential, hence the majority of the users that see a big difference are Epson users.

If we use Andrew's test as the measure, especially the balls, I don't think our results support that. My homemade profile through the Epson driver renders them pretty well, similar to your results. I suspect what we are seeing is profile differences. It looks to me like your left-hand black and white balls look better in the IP print.

Which brings me to my completely unsubstantiated opinion about IP: I suspect most users do a reasonable percentage of b&w prints. Part of the benefit of IP that is associated with Epson printers is the permanence improvement (uses less yellow ink).  I think Colorbyte puts more focus on their b&w pipeline and profiles. For example, the split tone tool is wonderful.

Some people swear by IP, I like it but it is definitely not for everyone.

Dave

[Ryan Mack]

I'd stick to actual photographs and not synthetic images. My understanding is the original LAB balls exceed the gamut of ProPhotoRGB which is itself much larger than the gamut your camera can capture. It's a bit like arguing which software does the best job with ultraviolet and infrared. If you do have problems with any real images then their customer support can figure out the issue with the profile and fix it for you.

For me the real value is the time savings from not making my own profiles and the consistency between printers and paper manufacturers. I do a lot of B&W and QTR was just a lot of extra work I'd rather avoid and the Canon and Epson B&W modes are so different to make consistency between printers a challenge.

[traderjay]

dchew - I will export the photo with banding as TIFF and reprint it again to see if there are banding issues. Nevertheless, there is still value with the 14 ball test because a standard benchmark image is the best way to gauge a software/hardware performance. But doing this on a consistent basis means an added step in my workflow and I am surprised that a $900 piece of software doesn't support RAW natively....

[Stngoldberg]

I fail to understand what anyone is saying on this thread.
To me the value if IP is the result of the printed image.
Like many others on this site, I work each image in PS or LR until the image on my calibrated screen depicts what I want that image to look like.
Then I depend on IP to deliver that exact image on whatever paper I am using at that time.
I find that IP almost always produces the desired result...end of story
Stanley

[Doug Gray]

You may be right, but that is a big conclusion to draw from one or two paper types in color. BTW, did you ever print those images you were struggling with as tiff outputs to see if the banding went away? I only ask because like your results with Bill's test balls, I was kinda shocked at first. As Schewe pointed out in that first thread I linked to, this type of test may not be all that relevant. Over the past two years, with several direct comparisons, I do not detect problems with color prints through IP. Of course, I may not know what to look for.

If we use Andrew's test as the measure, especially the balls, I don't think our results support that. My homemade profile through the Epson driver renders them pretty well, similar to your results. I suspect what we are seeing is profile differences. It looks to me like your left-hand black and white balls look better in the IP print.

Which brings me to my completely unsubstantiated opinion about IP: I suspect most users do a reasonable percentage of b&w prints. Part of the benefit of IP that is associated with Epson printers is the permanence improvement (uses less yellow ink).  I think Colorbyte puts more focus on their b&w pipeline and profiles. For example, the split tone tool is wonderful.

Some people swear by IP, I like it but it is definitely not for everyone.

Dave

The way the balls are rendered is entirely a property of the profiles and further, it's of limited utility.

The image is in ProPhoto. The balls are from Bill Atkinson and were not intended to be assigned to ProPhoto RGB. Atkinson intended them to be assigned to printer space. The purpose was to see what the smoothness of the printer RGB space is since that's critical in that visible, abrupt, banding indicates a rapid gradient change in the printer's channels and may demand a larger number of 3D LUTs to compensate. If banding in printer RGB space is abrupt enough it may exceed what profiling software can compensate for even with a really large number of patches. That's the reason Bill Atkinson created them.

In particular, the lower, far right "blue" ball is actually composed almost entirely of imaginary colors with a max luminance value ( L* ) of less than 1.  If the profile software designer chooses to favor preservation of luminance then rendering them as black is fine since the L* < 1.

The reason the "blue" ball appears as "blue" is because the conversion of these imaginary colors is well defined by the ICC when in the XYZ colorspace and is how conversion to the monitor, sRGB, or Adobe RGB spaces. These conversions bring the imaginary "blue" into a blue in the human gamut and raise the Luminance far above 0.

The I1Profiler's profiles also raise the Luminance when converting the imaginary blues and so creates less dissonance between what one sees on a monitor and what a print looks like. The main advantage is for people that get a bit carried away editing in ProPhoto and create colors significantly out of gamut or even out of the human gamut since one can't tell from the appearance on a monitor whether the actual RGB color is a real one or not, let alone a printable one.

That said, printing the "blue" ball as black is something Epson canned profiles typically do. It may or may not be related but the same Epson profiles I've examined are created incorrectly and scale the black point to the media black point. Effectively, this bakes in BPC whenever Rel. Col. printing intent is used. It also messes up softproofing as it doesn't show media blacks correctly.

[dchew]

Thank you Doug!

Dave

[traderjay]

Thank you all for the contributions and the in-depth technical info. Couldn't have got this anywhere else Doug Gray - I've to read your post a a few times and digest all the info! Never did I image that a simple test print image has so much color science behind it! Thats what make printing so much fun, the marriage between digital and reality.

I converted the photo that showed banding in the gradient into 16 bit TIFF and reprinted it in IP10 and the banding is still present. I reprinted the image three times using IP 10, Photoshop and DxO to make sure it wasn't a a fluke and the banding is present each and every time. Here is the image and the banding is extremely visible in the background and on my wife's face:



I've spent the past 10+ years trading the market while working in technical marketing in the semiconductor, imaging, machine vision and robotics field and takes a very critical approach at any marketing claims from companies and in this case I was let down by IP 10.

[traderjay]

I fail to understand what anyone is saying on this thread.
To me the value if IP is the result of the printed image.
Like many others on this site, I work each image in PS or LR until the image on my calibrated screen depicts what I want that image to look like.
Then I depend on IP to deliver that exact image on whatever paper I am using at that time.
I find that IP almost always produces the desired result...end of story
Stanley

Glad that its working for you and do mind sharing which printer you are using? In short I was hoping IP10 and my Canon Pro 1000 will yield incremental quality improvement over stock and had high hopes for the software. What I got was image degradation and quality issue totally contrary to the norm.

[Stngoldberg]

I use two Epson printers:
Epson 7880
Epson P 800
Stanley

[Ryan Mack]

I'd encourage you to contact their support email. They can probably get it sorted out.