Using an LCD screen as a 'digital negative' in alt-process contact prints

[AndrewBurns]

Earlier this year I attempted to use a modified laser cutter (fitted with a low-power 405nm UV laser) to make alternative process prints. I started with Cyanotypes as they presented a fairly fast and easy way to test my process. Long story short my laser exposure process worked but it was painfully slow and fairly low resolution, and I figured that the amount of work required to make it reasonably fast and high-quality just wasn't worth it. These are some of the best results I got (each of these A5 sized prints took more than 2 hours to make).



At the time some people recommended I look at the monochrome screens and UV exposure units used in resin 3D printers, and so I did some research and the concept seemed feasible. For some background, resin 3D printers use a tub of UV-sensitive polymer which is hardened by exposure to UV light that passes through an LCD screen. Mask images are displayed on the LCD screen to selectively harden only some of the resin, and multiple masks are shown one after the other to build a 3D part.

Before going any further, a number of people on this forum have already been using similar LCD screens in normal enlargers for silver printing with good results, there's a lot of detail to be had in the following thread: https://www.photrio.com/forum/threads/diy-31-megapixel-enlarger.197305/

The big difference with what I'm doing is using the screen for contact printing and with UV so I could make alt-process prints.

2D photo printing isn't exactly the same as 3D resin printing though, so the big questions I had were:

• How large of a print can I make (how big of a screen is available)
• How fast will the exposure time be
• What kind of resolution could I expect
• What sort of lifetime will the screen have

Most resin 3D printers are quite small, the largest have screens in the 14 to 15" diagonal range, so approximately the size of an A4 sheet of paper. Enough to make a good 8x10" size contact print, but anything larger would either require a different technique (like an enlarger) or tiling together multiple contact prints using some method of registration. I wanted to buy a complete module including the screen and UV lamp, and the biggest one I could find on aliexpress was 13.6" diagonal and 7k resolution.



This is the unit I bought, the aliexpress link is as follows: https://www.aliexpress.com/item/1005004111440256.html

The active area of the screen measures 298x165mm, it displays 6480x3600 pixels (roughly 23Mp) and when used for contact printing theoretically results in 552 dpi on the page, which should be more than enough. It's a monochrome screen which means each pixel can only show shades of grey, but this means it transmits UV light significantly better than an RGB screen would as there are no colour filters in the way.

The UV lamp is comprised of 84 405nm UV LED's with a large grid-shaped collimating lens in front of them. I haven't measured the power consumption but it's on the order of 200W+. It's extremely powerful and the output is well collimated, you can immediately feel your skin heating up when exposed to the light through a blank LCD screen (and keep in mind the screen already blocks nearly 90% of the UV light power).



My exposure time for traditional cyanotypes using this screen and light-source is around 4 minutes after adding a diffuser (more on that later). The rated life of the screen is >2000 hours when exposed to this much UV energy, which works out to something like 22,000 prints so I figure it's not really going to be a problem.

In order to actually drive the screen I needed to send it something over HDMI, so I bought a raspberry pi 4 computer and hooked it up. In the following photos I had an LED tracing pad behind the LCD screen so that I could see what it was doing.



And here's a view of the UV light shining through the screen with a strip along the middle set to clear while the rest of the screen is showing full black (plus a reflection from an overhead light).



Using the screen in this arrangement is simply a matter of displaying a picture of the negative on the computer in 'fullscreen' mode and then turning the UV light on. I'm currently performing this all manually, but in the future I'll write a program to automatically display the image and turn the light on for the correct length of time (or UV integration). I place my sensitised paper face-down on the LCD screen, put a piece of soft foam on top, then a piece of wood and some weights to keep the paper flat against the screen.



It's covered in the other Photrio thread I linked to at the start of this post, but sending an image to the screen isn't as simple as you might first think. Because it's a monochrome screen there aren't any RGB sub-pixels, so instead each R, G and B colour value actually controls the density of 3 physical pixels. The end result is that to send an image to the screen and have it look correct you first need to compress it by a factor of 3 in the horizontal axis and convert the compressed pixel greyscale values into colours. The resulting squashed and colourised image looks terrible on a normal screen but displays perfectly on the monochrome screen.



This squashed image is what gets sent to the screen (also inverted to be a negative and flipped) and this was the resulting print:



I think that this is a large enough post for now so I'll cover all of the other steps I've had to go through to get to that result along the way and I'll share the python script that I'm currently using to convert the image for display. The major topics I'll cover in future posts are:

• Diffusing the light source so that it gives even tones without making the image overly soft
• Linearising the cyanotype process to give a good output
• Limitations of the system and how I plan to work around them (primarily print size and dynamic range)

 

[Carnie Bob]

Not sure if this is the same application, but Mike Robiinson uses a computer screen which he photographs with his daguerrotype camera and makes stunning well detailed images.

 

[AndrewBurns]

Not sure if this is the same application, but Mike Robiinson uses a computer screen which he photographs with his daguerrotype camera and makes stunning well detailed images.

I think the possibilities of this technology are pretty exciting. It's only in the last few years that these screens have really progressed to the point where they could be a direct replacement for film in printing applications. A new 12k resolution LCD screen for example displays nearly 60 megapixels, so you could use it in an enlarger and make quite a large print or 'copy' it with a daguerrotype camera for example with huge amounts of detail. The limiting factor then is how well you scan your original negative (assuming you start with film), a drum scan would probably be required to get enough resolution to really make use of the screen's capabilities.

One could ask why use the screen at all if your aim is to match the quality and resolving power of MF or LF film when you could just use the film directly, but this kind of technology could allow you to take a shot with film, edit it digitally and then still print it using a traditional enlarger, or take a shot with a digital camera but print it using analogue methods. Or make alt-process prints without having to print transparencies with an inkjet or imagesetter.

 

[fgorga]

Andrew,

Thanks for the detailed information.

I am intrigued, however, I am not sure if pulling a project like this is within my skill set, especially the the software aspects.

Regards,

--- Frank

 

[AndrewBurns]

Ok so diffusing the light source. As I mentioned in my first post, the light source is well collimated from the factory, which is great for sharpness. I assume this is done for 3D printers because there could be some gap between the screen and the base of the resin container, and you want the print to still be sharp and high-resolution.

Unfortunately as you can see in the test print below with no diffusion, the light source is not very even, with a very clear grid-like pattern in the midtones. The manufacturer only specifies 90% illumination uniformity which I guess is fine for 3D printing but not for 2D photo printing!



My immediate thought was to add a layer of diffusion over the lenses to help even out the light hitting the LCD screen, however the only suitable material I had at the time was baking paper so I tried that. The result were even, however there was a clear loss of sharpness to the print which you can see below. (the photos look strange because the prints were hanging to dry at the time, but I can assure you they were rectangular).



So obviously some level of diffusion was good, but too much was bad. I went to a local camera and lighting shop and picked up some very light diffusion film, I then laser-cut a wooden frame to hold the film just above the lens array.



This was already a lot better, as you can see in the following test pattern, but there is still a visible grid pattern in the midtones.



I added a second layer of the same diffusion film over the first and the results got better again, still with acceptable sharpness.



I then set off on the task of linearising my process which I'll cover in a different post, but in my end result prints I could still see some non-uniformity in density that I didn't like.



I ended up getting a sheet of 'optisculpt' engineered diffusion film from this company: https://us.rosco.com/en/product/opti-sculpt

Specifically the 40 degree diffusion version. It's a much tougher and more robust film and it acts substantially more like a true gaussian blur effect vs. the film I had been using which was just milky/slightly opaque as you can see below.





I think the results I'm getting with this setup currently are better again, not 100% but close enough that I'm not going to get hung up on it for now. Here's the latent image of a print I tried making on inkjet paper. Unfortunately the image completely washed off during development but you can see from the latent image that the print is quite sharp and the smooth tone areas are pretty even.

 

[koraks]

This is all very interesting work and a great development in alt process printing. Being able to skip the iffy business with inkjet negatives is very valuable, I think.

 

[RalphLambrecht]

Earlier this year I attempted to use a modified laser cutter (fitted with a low-power 405nm UV laser) to make alternative process prints. I started with Cyanotypes as they presented a fairly fast and easy way to test my process. Long story short my laser exposure process worked but it was painfully slow and fairly low resolution, and I figured that the amount of work required to make it reasonably fast and high-quality just wasn't worth it. These are some of the best results I got (each of these A5 sized prints took more than 2 hours to make).



At the time some people recommended I look at the monochrome screens and UV exposure units used in resin 3D printers, and so I did some research and the concept seemed feasible. For some background, resin 3D printers use a tub of UV-sensitive polymer which is hardened by exposure to UV light that passes through an LCD screen. Mask images are displayed on the LCD screen to selectively harden only some of the resin, and multiple masks are shown one after the other to build a 3D part.

Before going any further, a number of people on this forum have already been using similar LCD screens in normal enlargers for silver printing with good results, there's a lot of detail to be had in the following thread: https://www.photrio.com/forum/threads/diy-31-megapixel-enlarger.197305/

The big difference with what I'm doing is using the screen for contact printing and with UV so I could make alt-process prints.

2D photo printing isn't exactly the same as 3D resin printing though, so the big questions I had were:

• How large of a print can I make (how big of a screen is available)
• How fast will the exposure time be
• What kind of resolution could I expect
• What sort of lifetime will the screen have

Most resin 3D printers are quite small, the largest have screens in the 14 to 15" diagonal range, so approximately the size of an A4 sheet of paper. Enough to make a good 8x10" size contact print, but anything larger would either require a different technique (like an enlarger) or tiling together multiple contact prints using some method of registration. I wanted to buy a complete module including the screen and UV lamp, and the biggest one I could find on aliexpress was 13.6" diagonal and 7k resolution.



This is the unit I bought, the aliexpress link is as follows: https://www.aliexpress.com/item/1005004111440256.html

The active area of the screen measures 298x165mm, it displays 6480x3600 pixels (roughly 23Mp) and when used for contact printing theoretically results in 552 dpi on the page, which should be more than enough. It's a monochrome screen which means each pixel can only show shades of grey, but this means it transmits UV light significantly better than an RGB screen would as there are no colour filters in the way.

The UV lamp is comprised of 84 405nm UV LED's with a large grid-shaped collimating lens in front of them. I haven't measured the power consumption but it's on the order of 200W+. It's extremely powerful and the output is well collimated, you can immediately feel your skin heating up when exposed to the light through a blank LCD screen (and keep in mind the screen already blocks nearly 90% of the UV light power).



My exposure time for traditional cyanotypes using this screen and light-source is around 4 minutes after adding a diffuser (more on that later). The rated life of the screen is >2000 hours when exposed to this much UV energy, which works out to something like 22,000 prints so I figure it's not really going to be a problem.

In order to actually drive the screen I needed to send it something over HDMI, so I bought a raspberry pi 4 computer and hooked it up. In the following photos I had an LED tracing pad behind the LCD screen so that I could see what it was doing.



And here's a view of the UV light shining through the screen with a strip along the middle set to clear while the rest of the screen is showing full black (plus a reflection from an overhead light).



Using the screen in this arrangement is simply a matter of displaying a picture of the negative on the computer in 'fullscreen' mode and then turning the UV light on. I'm currently performing this all manually, but in the future I'll write a program to automatically display the image and turn the light on for the correct length of time (or UV integration). I place my sensitised paper face-down on the LCD screen, put a piece of soft foam on top, then a piece of wood and some weights to keep the paper flat against the screen.



It's covered in the other Photrio thread I linked to at the start of this post, but sending an image to the screen isn't as simple as you might first think. Because it's a monochrome screen there aren't any RGB sub-pixels, so instead each R, G and B colour value actually controls the density of 3 physical pixels. The end result is that to send an image to the screen and have it look correct you first need to compress it by a factor of 3 in the horizontal axis and convert the compressed pixel greyscale values into colours. The resulting squashed and colourised image looks terrible on a normal screen but displays perfectly on the monochrome screen.



This squashed image is what gets sent to the screen (also inverted to be a negative and flipped) and this was the resulting print:



I think that this is a large enough post for now so I'll cover all of the other steps I've had to go through to get to that result along the way and I'll share the python script that I'm currently using to convert the image for display. The major topics I'll cover in future posts are:

• Diffusing the light source so that it gives even tones without making the image overly soft
• Linearising the cyanotype process to give a good output
• Limitations of the system and how I plan to work around them (primarily print size and dynamic range)

I tried using an iPad to do what you are suggesting, but I haven't been able to. My problem was that I couldn't dim the iPad screen enough to get reasonable exposure times.

 

[calebarchie]

How funny, I just started working on something similar today but for a different type of printing process. I'm using the 12K LCD from my printer which has rectangular pixels that are slightly annoying to deal with. How much did the engineered diffuser set you back?

C

 

[AndrewBurns]

How funny, I just started working on something similar today but for a different type of printing process. I'm using the 12K LCD from my printer which has rectangular pixels that are slightly annoying to deal with. How much did the engineered diffuser set you back?

C

It was ~$116 AUD for a 61x51cm sheet. The stuff I got is 40 degrees diffusion angle from one side and if you flip it over it's 60 degrees.

 

[calebarchie]

It was ~$116 AUD for a 61x51cm sheet. The stuff I got is 40 degrees diffusion angle from one side and if you flip it over it's 60 degrees.

Thanks! Sounds like you may have some to spare? DM me if so I may be interested, just across the ditch.

 

[AndrewBurns]

Well I can't call the diffusion a solved problem yet, it wasn't very obvious with the test print I had been making but this one really shows how obvious the grid illumination pattern is in large solid areas of light-grey.



The print is just out of the developing bath so I expect it will sharpen up a fair bit when the paper dries.

Kinda frustrating, I'll have to try more diffusion again. I can flip my current sheet of diffusion film around to get 60 degrees of diffusion angle rather than 40 and I might have to start stacking multiple layers.

 

[koraks]

I'll have to try more diffusion again

Can't you just increase the distance between the light source and the print/contact frame? That would solve your uneven illumination problem very effectively. Exposure efficiency would drop a little of course, but you have some power to spare.

 

[AndrewBurns]

Yep that's also a possibility I'm thinking of. I'm just hesitant to do it as then I'd need to make something to tack onto the nice pre-made module I bought which is extra work but as you say it should be quite effective. I want to keep the whole thing somewhat compact as I'm imagining in the future mounting the LCD and light unit to a board that I can physically pick up and put down onto prints, enabling printing larger sheets with the use of multiple registration pins.

 

[Carnie Bob]

To the OP- when I was working in professional photo labs in the late 70's , 80's and 90's I always dreamed of a system of being able to accurately focus, sharp edge to edge prints, dodging, burning , contrast , density and colour correction to be handled much faster and easier than what I learned making prints Colour and Black and white on enlargers and using all kinds of physical tricks to get what I want. Fast forward to today, I. am 71 years of age and am using PS and different devices to to exactly that with the skill and accuracy of the 80's. In fact I have some of the same equipment and working methods mixed in with current day technology.
There is a future for printing that I envision but frankly not willing at my age to invest and go down another 10 year wormhole and part of what you are doing is exciting and very encouraging for me to see. I have a young apprentice who is learning all the tricks that I know and hopefully people like you figure out a new printing direction .

What do I see as the future... well mixing gravure printing, various photo processes on computerized flatbed exposing devices that are XY stable to provide multilayered print that will last for centuries. Will there still be those of us using enlargers and analogue methods, yes of course, its what I know best but I hope to see her and people like you expand the parameters..
One only has to look at what Salto tried to do in Japan with PT Pd prints to understand this new approach is totally possible.
Who would have thought young photographers are buying printing presses and doing direct to polymer plates impressions on paper, not me but they are and this is exciting .

 

[AndrewBurns]

Bob, I'm old enough to have shot film as a child before digital really existed but young enough to have only ever taken photography seriously after digital had taken over, and so I'm only used to editing photos using digital tools. I've since discovered that I love taking photos using a large format camera and also the involvement of hand-made prints, but I can't give up the power and flexibility of digital photo editing (which also makes it easier to share my photos with people).

I think this kind of hybrid workflow is the best balance for me, I get the tactile engagement of shooting a scene with LF film, but then I scan and edit digitally. I do have a decent inkjet printer but I've always hated using it, so I'm excited to be working on a way of printing my digitized files in a way that is still genuinely involved and hand-made.

Like digital photo editing I think it also makes some techniques that would be possible but quite time consuming using more traditional methods easier and faster. For example if I wanted to make an alternative process print that required building up multiple different layers (e.g. multi-tone or colour prints) then I don't need to print out multiple digital negative transparencies or shoot multiple sheets of film, the screen can print as many 'negatives' as I send to it. I can also increase the tonal range of the screen by using software to automatically break an image up into different copies with slightly shifted exposure, which get automatically exposed one after the other onto the same print, kinda like how HDR photos work but in printing instead.

 

[grain elevator]

Yep that's also a possibility I'm thinking of. I'm just hesitant to do it as then I'd need to make something to tack onto the nice pre-made module I bought which is extra work but as you say it should be quite effective. I want to keep the whole thing somewhat compact as I'm imagining in the future mounting the LCD and light unit to a board that I can physically pick up and put down onto prints, enabling printing larger sheets with the use of multiple registration pins.

I don't think you'll get around that. Think of it this way: light needs to get from a lighter field of the array at least all the way across a neighbouring darker field, or more if you take into consideration that it's not just a checkerboard pattern but you'll need to equalize differences of not directly neighbouring fields. For the light to equal out over these distances, I'd expect you need a distance between diffuser and paper in the same order of magnitude. It will still be much flatter than the alternatives.

 

[AndrewBurns]

Yes there's currently about 1 inch of air-gap between the diffuser film and the back face of the LCD screen, which is probably on the same order of magnitude as the 'cell' size of the collimating lens array, but yes I agree the light field evenness would be improved if this was made larger. I assume this would not be the case without the diffusing film as if the collimating lenses are working correctly the light wouldn't be diverging much over even large distances.

Another options I'm considering is a flat-field correction of the image shown on the screen, effectively adding a grid-shaped mask to the image to compensate for the variation in brightness of the light field: https://en.wikipedia.org/wiki/Flat-field_correction

This should be possible as long as the LCD and lens array don't move relative to each other (which is fine as they're both part of the same assembly). In theory I would make a solid 'grey' print and then scan it to get a measure of the light field non-uniformity, which I could then apply to the image to be displayed, however in practice I'm sure it could be challenging and might require some iterations and tweaking to get it to work.

I found a python implementation of the technique here: https://stackoverflow.com/questions...gnetting-template-from-image-in-opencv-python

 

[calebarchie]

This project reminds me of a similar one on Hackaday from a long while back. The fella made a module with a UV laser and polygon mirror but the project was abandoned. It probably did not offer enough speed benefits exposing line by line, I do know there are much faster emulsions nowadays at least for cyanotypes.

FFC seems to be the way to go, good luck with it!

 

[koraks]

The fella made a module with a UV laser and polygon mirror but the project was abandoned.

@AndrewBurns that must sound familiar to you, I guess! Well, minus the polygon mirror. Are you planning to write about your laser experiments here?

 

[Esso_71]

This is super interesting, thanks Andrew! Can you buy those higher resolution B/W LCD screens without an UV source as plug-and-play without the need of an adapter to a PC or so? I still use my trusty 30 years old skin-tanner as an UV source and wonder if I can "just" place an LCD in between?

Best,
Stefan

 

[koraks]

LCD screens without an UV source as plug-and-play without the need of an adapter to a PC or so?

There needs to be a computer of some sort that generates the image for the LCD. Most LCD modules you'll find are bare modules without even a display driver IC (basically useless as such), or they come with a driver circuit that accepts something like HDMI input. You'll be hard pressed to find a bare LCD that would work in this application that also functions as a stand-alone module, displaying files from an SD card or something like that. Refer to this thread for more info as it's conceptually similar: https://www.photrio.com/forum/threads/diy-31-megapixel-enlarger.197305/

 

[AndrewBurns]

@AndrewBurns that must sound familiar to you, I guess! Well, minus the polygon mirror. Are you planning to write about your laser experiments here?

Koraks, yes I mentioned it in my first post very briefly but I should make a different post about that sometime! There's just so much to do and talk about and so little time to do it in. Suffice to say that I looked at lasers, it worked and could be made to work quite well if you were willing to put a lot of time and money into it. But I'm already getting much better results with this LCD screen with at less time and money investment then I think the laser option would have needed, mostly because you can buy all of the hard engineering problems solved for you off the shelf, in the form of the screen.

This is super interesting, thanks Andrew! Can you buy those higher resolution B/W LCD screens without an UV source as plug-and-play without the need of an adapter to a PC or so? I still use my trusty 30 years old skin-tanner as an UV source and wonder if I can "just" place an LCD in between?

Best,
Stefan

As Koraks has mentioned you can buy just the screen but most sellers will also sell you an adapter board to be able to connect to the screen via a HDMI cable which you would also want. With that you can technically drive the screen with anything able to send a 4k image over HDMI, but you need to do some 'work' to the image before you send it if you want it to look correct on the screen. You can get slightly larger ones than I have, and also many smaller ones with higher resolutions. I got mine from Aliexpress, lots of options on there if you search for something like "mono LCD".

I set up my DSLR over the light box and screen tonight to have a quick go at capturing an image of the light field uniformity. I used a single piece of plain printer paper on the screen to diffuse the light so that the camera could capture it (otherwise it's still far too collimated for the camera to get an even photo of).



Here's a shot of the paper with my phone.


Here's a shot from the DSLR, with some processing to make the pattern stand out more. Obviously the setup needs some work, primarily replacing the sheet of paper with something more even, like translucent acrylic, but it looks feasible.

 

[AndrewBurns]

I know it's been a little while, have been super busy with other things in life so haven't had much time to work on this. I have been cleaning up my image squashing code though so I thought I'd share it here. I'll only share a portion of it though as the bulk of it still isn't really ready for 'prime time'.

As I mentioned previously, the LCD screen 'cheats' a bit by mapping entire pixels to each of the RGB colour channels that get sent over HDMI, this saves a significant amount of bandwidth and means you could control a 12k resolution screen from a computer only capable of 4k video output (like a Raspberry pi).

Here's a diagram of what I mean:


Basically every 3 grey pixels in the horizontal axis become a single colour pixel, and when this gets sent to the screen it gets 'unpacked' back into the original 3 grey pixels.

The code for doing the squashing is quite simple, written in python and using the OpenCV image processing library. This code chunk assumes it's already getting a greyscale image

import cv2
import numpy

def Image_squash(image_in):
    # Get image shape
    image_y, image_x = image_in.shape
    
    # Make empty image of final shape
    image_out = numpy.zeros((image_y,int(image_x/3),3))
    
    # Compress width of image into colour space
    # For every three x-axis pixels, compress them into a single colour
    for y in range(0,image_y):
        for x in range(0,int(image_x/3)):
            image_out[y,x,0] = image_in[y,(x*3)+2] # Blue channel
            image_out[y,x,1] = image_in[y,(x*3)+1] # Green channel
            image_out[y,x,2] = image_in[y,(x*3)] # Red channel
    
    return image_out

You also have to invert the image so that it becomes a negative and flip it so that it prints correctly, as well as pad the edges so that it fills the entire screen. In practice this turns an image that looks like this:



Into one that looks like this:



And that second squashed image actually looks correct on the screen!

 

[koraks]

The code for doing the squashing is quite simple, written in python and using the OpenCV image processing library.

Man, that makes me chuckle; I'm working on some embedded C and C++ at the moment and was troubleshooting a startup problem with a microcontroller just now. It takes a couple of hundred lines of machine instructions just to have the device start executing the first line of user code, LOL!
Those higher-level scripting languages are so nicely convenient. It's just wonderful.

 

[AndrewBurns]

Yep I'm basically a mechanical engineer, my aim is to know just enough about coding to do what I want to do, so languages like python are great. Not fast, but easy to use and with a lot of pre-made libraries.

I had a quick attempt at applying the flat-field correction to my light source, no time to do it properly yet but results are promising. This is using a piece of plain paper as a diffuser which is why there's so much texture to the image.

Uncorrected light field on the left, corrected on the right.


It doesn't look like much of an improvement until you look at the heavy vignette around the edges of the uncorrected screen, which is nearly gone on the corrected side.

I think I'm getting my correction image flipped or rotated somewhere along the toolchain I'm using which is causing the correction to be poor, but once I have time to figure it out I think I can get it to work pretty well.