Some tests with the ET-8550

[stratman]

light inks

I thought a benefit from these light inks were to smooth color gradients and decrease banding / stepping in transitions.

 

[Ink stained Fingers]

I didn't think there was that much on the internet interesting enough to print.

What ink is used the most and what do you calculate the amount used so far?

I don't like to read the news on the display and rather print a lot - probably 2000 sheets per month, this mainly on a L805 at this time to use up no-name ink and left-overs from all over the world, I'm settling down to the Epson 106 inks only for dye inks.
I'm not tracking ink usage on the ET-8550 , I just topped off the containers with 106 ink - the starter set was the 114 ink set, they don't vary much by their gamuts, mixing does not cause much of a color shift.

 

[Ink stained Fingers]

That’s why I use this printer with the four dye inks. It’s good enough for the job. On formats like A3+ or larger there may be differences with certain prints like portraits with a lot of subtle skin tones or with a lot of greens

I thought a benefit from these light inks were to smooth color gradients and decrease banding / stepping in transitions.

That's what I read a lot as well but there is no real proof, color perception is more complex.

There is the technical side - with RGB 3x8bits you can create 16 Million different colors - and now one simple question - how many different colors are actually printed by a 4 or 6 or 10 color printer - what is the driver doing ?
This is one of the well kept secrets by Canon or Epson - I never have seen any info to this extend .

And now the next question - how many colors can your vision system actually differentiate - your read about vastly differing numbers like a few thousend or ten thousends or some hundredthousends or ? Quite some of that has been tested over decades by the ICC - color consortium - trying to measure and standardize visual color differences - this with various generations of a deltaE value. It's as always - it depends - on the overall brightness , the hue, the saturation which makes it so difficult to press it into one number.

Small color variations - those subtle ones - are actually not visible unless they are larger than a deltaE of 1 under best conditions - or 2 or 3 closer to a real life situation.

Just take a grayscale 'continuous' ramp, is it continuous when RGB changes from 1 to 2 or 34 to 35 or 254 to 255 ? It's a ramp with 1 bit steps but just perceived as continuous , a change of 1 bit - or 0.4% - is just not visible, and you can do the same with color ramps of various types - you don't see 1 bit differences., not with an ET-8550 or with an WF2010W with 3pl droplets. If you see banding or alike there is a problem - mostly with a color profile, or some prior excessive manipulations with an image editor.
So I'm very sceptical about those claims about 'subtle color changes' unless clearly specified what is meant, it could be that saturated colors are on the edge or out of gamut which let you loose detail definition, but that's not something you can cure with light inks.
And again - we don't even know how many different colors are Epson or Canon printers capable to print technically.

 

[stratman]

I don't like to read the news on the display and rather print a lot

I understand. Certain things are preferable on printed paper.

Same with texting versus calling and talking.

 

[maximilian59]

You can’t create 16 million colors with the grid. There are a lot of mathematical points in a cube. A color that cannot be seen is no color per definition. The human eye can distinguish 2.000.000 colors say the one, up to 20.000.000 say others. Nobody has started yet to check the figures. It would take to long. Most I found that for the normal viewer it is more the 2.000.000 colors for untrained eyes. Seeing colors is not discret in little cubes and they are not linear in the room. The grid describes locations, whether there is a color or not.
Colors itself have nothing to do with bits. That is only trying to transform something continuous in divided parts. There are a lot of areas with bits, but no visible colors. If you look at the so called horse shoe you see big differences in distances from white to fully saturated. With 8 bit you divide the distance by 256. What is with all the colors in between? Are they not existing because they are not hit by the coordinates? How fine do you have to go, to say this is a color or not?
All these things with coordinates are mathematical models to describe something. For example you will not get white, even you print nothing, depending on the paper. Color perception is something really different to colors in bits. The one sees differences, the other not.
The more possibilities the device gives you, the more possibilities exist, whether you need them or not.
That is my opinion and what I can see. And if I look at prints from a P5000 I am always impressed, also my Pro-1000 gives colors which are astonishing.

 

[Ink stained Fingers]

Let's not split hairs - or colors, here are lots of color values - for the RGB 3x8 bit scenario ,
that's 16.777.216 colors in total.

We don't know how many different colors are actually printed by the driver , that's the technical part of it, and we know even less how many colors we acctually can visually separate, it's def. less than millions of colors.

A color that cannot be seen is no color per definition.

No, not necessarily - such color might be substituted by another color which creates a visible appearance, but this does not help to assess possible benefits of light inks - I think that light links are not needed anymore in Epson printers with 1.5pl droplets, a variety of A4 and A3 models demonstrate this.

And if I look at prints from a P5000 I am always impressed, also my Pro-1000 gives colors which are astonishing.

Yes, those printers are tuned for a wider gamut - with more inks , but that's most likely not related to light/photo inks.

 

[maximilian59]

I think we talking about different things.
Let’s say all visible colors are in a sphere. What’s not visible is not interesting. This sphere is in a cube, the RGB system. You can describe about 25 % more locations with coordinates, but they will be never part of the sphere. Inside the sphere you can describe locations with coordinates, but is every coordinate a different color? Next you go to a 64 bit system and are there more colors now?
If we measure colors you get values with two two digit es behind the comma. With more digits are there more colors now or what are we measuring. If you measure five times and make the average is this the correct color or just part a coincidence of measurements?
As far as I know from color theory colors are a continuous spectrum of wavelengths and nobody knows how fine one can go. You can stretch it and stretch it and it seems there will be no missing spots.
If I stretch a 8 bit color gradient, you get missing colors and artifacts. With 16 bit it is much better. So if light colors give you more bits than it’s better. Whether you see it every time is someth different. It’s the same with 8 bit files and 16 bit files.

 

[Ink stained Fingers]

I think we talking about different things.

If I stretch a 8 bit color gradient, you get missing colors and artifacts

Yes, but that's not something the printer does, you stretch the gradient, and you see the steps and artifacts in the printout, and yes - it is better to do photo/image editing in 16 bit mode for that reason - specifically if you work in a wider color space , but that's all prior to printing.
If I print an 8 bit grayscale continuous ramp I don't see any steps and gray level changes which relate to the 1 bit increments , it appears to be continuous, and that's the same with 1 bit color changes, you don't see them - and if you see artifaccts then the problems are caused prior to the print - typically editing issues or a profile screw up in some cases . I must admit I don't really know what to do with these 'subtle changes' - I either don't see them or I see them, but then they are not 'subtle ' anymore. My threshold for visibility may be lower or higher than by some other people which may complicate the discussion slightly, but it remains important to send decent data to the printer in the first place , I don't like those automatic enhancement/correction features in the driver like edge smoothing, image/contrast/color corrections. I only can say that I'm happy (enough) with the color output of the ET-8550 in this respect.

I just don't think that a printer with light colors generates better output just because it is using these light inks , I still see the need to reduce the droplet visibility when larger droplet volumes are involved on large format printers, and the gamut gain with additional colors is another, nevertheless interesting subject.

 

[The Hat]

I found that for the normal viewer it is more the 2.000.000 colors for untrained eyes. Seeing colors is not discret in little cubes and they are not linear in the room. The grid describes locations, whether there is a color or not.

My threshold for visibility may be lower or higher than by some other people which may complicate the discussion slightly

When it comes to colours, we’re only as good as our eyes think we are, whether its 4 colours or 40,000 that we’re looking at, we are so easily bamboozled and can swear blind that we see more.. But in fact we don’t see the half of them..

These are only examples of how things are not what they perceive..



P.S. No printer was used or harmed in making these imagines..

 

[Ink stained Fingers]

I did another test to compare printouts and to see whether and which differences might be visible , I printed the Fujifilm test image on the ET-8550, the L1800 with 6 inks and a WF-2010W with 3 or 4 inks.



I'm printing this image with the Epson native resolution of 720 dpi for these printers which results in a print size of 116 x 82 mm. I'm focussing onto the small shot with the foto models which print at 17 x 11 mm and do some macro shots to get down to the pixels and droplets.



I'm printing with the high and best quality settings for the Epson Ultra Glossy - UG - paper on the ET-8550 and get these prints



I don't see much of a difference here. And this is the same print on an L1800 printer - with light links, the driver does not offer a 'best' quality option, just standard and high settings for the ultraglossy paper, this printer runs as well with 1.5 pl droplets as the ET-8550 does .


The WF-2010W runs with 4 inks in various combinations -there are printers and combos with prigment inks and models with dye inks offered by Epson, or a mix with a pigment black and dye color inks. The driver for all these models does not use the black ink - whatever it is - if you select a glossy photo paper, it is mixing the inks to get a not so black blue-violett color, it does not really look pleasant, if you use the matte or inkjet paper setting the driver uses the black ink channel and you get decent print output. The larger droplets of 3 pl are visible in the macro shots but not in regular prints, we could go down to the arcsminutes for the limit of visibility of fine details - something like 0.15 mm in 25cm distance - the actual droplets are smaller - but you see as well other numbers quoted on the internet. It does not matter much - you most likely would need glasses for close-up viewing anyway for a 25 cm viewing distance.



Since a printer like this WF-2010W does perform quite well for photo printing Epson is restricting the use of some options intentionally, the WF-2010W prints borderless, there is a piece of absorber foam at the bottom of the print mechanism to catch the overspray from borderless printing. That's fine to this point, I'm not talking about the low print speed for photo printing since there are only 3x59 nozzles for the CMY colors.
There are Ecotank variations with this print mechanism, some don't let you print borderless at all, the absorber is not there, or you just can print borderless on 10x15cm, not A4, the absorber strip has only half of the length.
And there are various ink combinations - all 4 inks are dye inks but the black is still not used for glossy paper, or some models use a pigment black and 3 dye inks or combos use 4 pigment inks like the WF-2010W and other WF.... models more tuned for home office tasks.
If you do refill or use 4 dye inks on a printer version which supports borderless on A4 - the L310 did not but the L382 did - you have a pretty good photo printer for occasional use , the graininess is not visible, and the gamut of a WF2010W is pretty close to the photo printers. But Epson would not like you to know this and rather sells higher priced 'photo printers' to you. (But again - the printing speed is slow which should be o.k. for occasional use).
I'm running a WF2010W mainly for ink testing , with a rather wide set of refill cartridges - pigment and dye - Canon - Epson - 3rd party - China etc.

The L1800 as the 1400 Ecotank A3 photo printer version uses 6 inks - with light C and M and one black ink channel only, the droplet size is as well 1.5 pl for the smallest ones. But the user does not have any control over the use of different droplet sizes via the driver, it is hidden for him.

The actual print does not look very much different to those prints from the ET-8550