All in all modern epson printers produce amazing results. But I seem to have found a weakness which I'm finding most frustrating. I am one of those crazy people who hopes to sell prints so I have to be a little fussy. I worked in a professional color processing lab for some five years so have enough knowledge of color matching to be dangerous.
The problem is this – when yellow (particularly warm yellow) blends into black shadow - a purple band appears at the boundary. This appear to be a fundamental blending problem in some models.

Shadows in photos often have a real blue cast from the sky but this isn't the case here. The subject is a flower and there is a slight red cast from light diffusing thru the petal – not blue purple. On the left is a portion of the digital image (powershot G2) right is a scan of a print on my 1160. No effort has been made to color match the two. Click on the images to enlarge and you should see the problem. A print on a canon s400 does not have this effect.
My printer has a home-made CIS and I run “generations” G4 inkset. I also wrote the ICM/ICC profile software I use.
To rule out ink or profiling to be the problem I printed a test on a 2000p using OEM ink and no profile. This also had a purple band.
It could still be a problem with epson heads and pigment.

I tried to produce a software generated image with the same effect, this would rule out sky-light etc. This image works to some extent – not a clear as the G2 image but the purple is there to some extent – OK not much I confess. It is more obvious on the print than the scan.

I have considered defecting and buying a canon s9000 (avoids the chipped ink cart nuisance too) but the pigment inksets are not ready for it – having over 3000 nozzles to clog is a bit scary too.

Ultimately I bought a 2100P. It is much better but some banding is still present, this time yellow and in a different place.

Some feedback.

John Houghton wrote:

Eddie, I tried a print of your flower on a 1290 with Print-Rite cartridges

installed and I could see no evidence of a purple band like the one in your

sample. I assume you were viewing in daylight. My inks exhibit a magenta

shift due to metamerism in artificial light..

Regards

John

Thanx, this is useful data point. A 1290 using pigment might be different. One possibility is that yellow pigments don't work down to the 4 pico litre drop sizes. Since yellow never seems to clog I doubt it but who knows?

And

Howdy eddie,

Had a similar problem recently on an EX using lyson fotonic inks. Photo was

sand going into shadow (so kind of beige colour going to black) produced a

magenta band.

Snip snip...

I'm just waiting to get a 2100 so I'll be looking out for a similar effect.

Geth

This is just pretty image I made with my test pattern generating program. Makes nice wallpaper :)

The rest of the site is here.
Email address is here