Color vs grayscale digital negatives

[Andrew O'Neill]

Seven cartridges in my P400 are PK, and one is yellow. I kept yellow because it blocks UV very well.

 

[NedL]

Post number 40 in this thread seems to be pretty important.

I've recently been making the gradient based on what jisner calls "Color Fill Layer" ( a gradient through the color space with fixed hue, with both saturation and value changing through the gradient. From light to dark: saturation increases and value decreases simultaneously. ), and a couple of days ago I switched to what jisner calls "ColorBlocker" ( a gradient based on a fixed hue, from light to dark: value fixed at 100 saturation changing from 0 to 100, followed by saturation fixed at 100 and value decreasing ). There are some interesting consequences in the resulting print that are not obvious from this thread or from just thinking about it. I'm going to list a few things I noticed here, in case anyone is interested. I will call the Color Fill gradient method 1 and the ColorBlocker gradient method 2.

Before I do, I need to mention that I'm using a cheap HP office inkjet and the printer decides what mixture of inks to use for each color, and how much ink to use. I have no control over it and no way to know what combination it is choosing. It chooses when to mix in black inks ( and it makes that choice differently for different colors ) and the black ink does not block UV as well as some of the colored inks. I'm making sepia prints and VDB prints. I'm not going to try to explain why these differences are happening, just some practical observations that matter to making a print.

• Method 1 gives a visually smoother gradient on the final print. Method 2 has a noticeable change at the point where the type of gradient changes ( from 100% saturation to 100% value ), and the two regions of the gradient behave very differently. The part of the gradient where value is held at 100% contains visible changes in tone on the print, but the part of the gradient where saturation is held at 100% has much less difference between tones, changing more slowly, and with parts of the gradient being indistinguishable on the print. A practical consequence of this is that contrast changes in the original positive image has unpredictable effects in the contrast of the midtones of the negative, depending on where the transition of gradient types happens. I have an example in front of me where gamma=1.8 original positive appears to have less contrast than gamma=1.5.

• With my printer, method 2 makes the print look much more grainy/granulated across all but the darkest tones in the final print. For whatever reason, the part of the gradient that has value fixed at 100% makes a mixture of inks that causes very obvious graininess. I don't know if it is black inks being mixed in or just a mixture of inks with different UV blocking characteristics, but method 1 produces much smoother looking tones while all but the darkest tones have obvious "grain" using method 2. On my prints this amount of grain is not acceptable.

• This one is interesting! There are substantial color differences in the final print. Using method 1, all the tones appear to be the same hue on the final print, just with changes of value ( in the artist's meaning of "value" ). With method 2 the very darkest tones ( where the negative is clear ) have that same hue, but then all dark tones are a warmer color, abruptly switching back again to the original cool tones . The effect is different on different papers ( that print with different colors ) but visible on all of them, and in the same transitions points on all papers. The transition from warm color back to the original cool color happens while the gradient is still in the part where value=100% and saturation passes through around 50%. I suspect different inks in the mixture allow different parts of the UV spectrum through, and even though there is still a gradient in (artist's) value, the color is shifting because the reduction of silver varies with spectrum of UV getting through the negative. This also shows that my printer has multiple "thresholds" in which inks are used.

Kind of humorous thing is that when I read Peter Mrhar's page the other day, I thought I was setting up the gradient wrong, so I changed what I was doing. Now I'll go back. Thanks for post #40, that was very helpful.

 

[nmp]

Post number 40 in this thread seems to be pretty important.

I've recently been making the gradient based on what jisner calls "Color Fill Layer" ( a gradient through the color space with fixed hue, with both saturation and value changing through the gradient. From light to dark: saturation increases and value decreases simultaneously. ), and a couple of days ago I switched to what jisner calls "ColorBlocker" ( a gradient based on a fixed hue, from light to dark: value fixed at 100 saturation changing from 0 to 100, followed by saturation fixed at 100 and value decreasing ). There are some interesting consequences in the resulting print that are not obvious from this thread or from just thinking about it. I'm going to list a few things I noticed here, in case anyone is interested. I will call the Color Fill gradient method 1 and the ColorBlocker gradient method 2.

Before I do, I need to mention that I'm using a cheap HP office inkjet and the printer decides what mixture of inks to use for each color, and how much ink to use. I have no control over it and no way to know what combination it is choosing. It chooses when to mix in black inks ( and it makes that choice differently for different colors ) and the black ink does not block UV as well as some of the colored inks. I'm making sepia prints and VDB prints. I'm not going to try to explain why these differences are happening, just some practical observations that matter to making a print.

• Method 1 gives a visually smoother gradient on the final print. Method 2 has a noticeable change at the point where the type of gradient changes ( from 100% saturation to 100% value ), and the two regions of the gradient behave very differently. The part of the gradient where value is held at 100% contains visible changes in tone on the print, but the part of the gradient where saturation is held at 100% has much less difference between tones, changing more slowly, and with parts of the gradient being indistinguishable on the print. A practical consequence of this is that contrast changes in the original positive image has unpredictable effects in the contrast of the midtones of the negative, depending on where the transition of gradient types happens. I have an example in front of me where gamma=1.8 original positive appears to have less contrast than gamma=1.5.

• With my printer, method 2 makes the print look much more grainy/granulated across all but the darkest tones in the final print. For whatever reason, the part of the gradient that has value fixed at 100% makes a mixture of inks that causes very obvious graininess. I don't know if it is black inks being mixed in or just a mixture of inks with different UV blocking characteristics, but method 1 produces much smoother looking tones while all but the darkest tones have obvious "grain" using method 2. On my prints this amount of grain is not acceptable.

• This one is interesting! There are substantial color differences in the final print. Using method 1, all the tones appear to be the same hue on the final print, just with changes of value ( in the artist's meaning of "value" ). With method 2 the very darkest tones ( where the negative is clear ) have that same hue, but then all dark tones are a warmer color, abruptly switching back again to the original cool tones . The effect is different on different papers ( that print with different colors ) but visible on all of them, and in the same transitions points on all papers. The transition from warm color back to the original cool color happens while the gradient is still in the part where value=100% and saturation passes through around 50%. I suspect different inks in the mixture allow different parts of the UV spectrum through, and even though there is still a gradient in (artist's) value, the color is shifting because the reduction of silver varies with spectrum of UV getting through the negative. This also shows that my printer has multiple "thresholds" in which inks are used.

Kind of humorous thing is that when I read Peter Mrhar's page the other day, I thought I was setting up the gradient wrong, so I changed what I was doing. Now I'll go back. Thanks for post #40, that was very helpful.

Thanks Ned for sharing your results.

So your conclusion is don't mess with the Mehar ColorBlocker thingy, right? I never did it and never understood the need for it other than it's just another way to do it - kind of taking a longer detour when the straight road home works just fine.

:Niranjan.

 

[revdoc]

My understanding of the colour blocker step is that tries a large number of colour combinations (more or less randomly) for one that's better than using just one colour by itself. The HSV grid is just one possible input for this. It's a fairly arbitrary process that only has a limited, but non-zero, chance of success. In fact I think I mentioned the luck aspect of it in earlier post.

In case I haven't mentioned it before, this paper seems to have preceded Peter Mrhar's usage of this technique:

https://web.archive.org/web/2014021...ages/RNP/quick_guide_to_making_digital_ne.htm

 

[NedL]

Thanks Ned for sharing your results.

So your conclusion is don't mess with the Mehar ColorBlocker thingy, right? I never did it and never understood the need for it other than it's just another way to do it - kind of taking a longer detour when the straight road home works just fine.

:Niranjan.

I suspect the EDN thing will work OK depending on how the points on the correction curve are selected ( I don't know if the software interpolates or if it selects the "nearest match" to find those points ). But using that kind of gradient seems to introduce unnecessary discontinuities and parts of the gradient that are less useful. The 15 points on the correction curve need to make a more extreme correction. With the color I've been using ( hue 110 ) on my printer, a "perfect" curve would be nearly vertical in some parts of the mid-tones. I really don't mean to criticize Mr. Mrhar, who after all is making his approach freely available and trying to make it as easy to use as possible, which hopefully makes printing attractive and accessible to more people! That big picture is more important than the type of gradient