DIY 31 Megapixel Enlarger

[koraks]

Brilliant! Are you going to do color with the machine as well?

 

[radiant]

Brilliant! Are you going to do color with the machine as well?

Do not provoke me

But seriously; after calibration the color printing would be as easy as B&W printing. I don't have color printing equipment, most imporantly the jobo drum machine so probably not in short tems.

 

[koraks]

The jobo thing isn't really necessary; you could give it a try with trays on room temperature first.

The main challenge is to create a color profile and/or calibrate the thing. I'd be tempted to try the method put forth by Calvin Grier in his calibration ebook series. This seems to be an application where his approach would work exceedingly well (without the need for separations btw).

 

[avandesande]

The problem isn't necessarily bandwidth, it is having actual chips to drive the y axis at that refresh rate. There is no specific technical issue with making a chip but there is nothing like it and I am guessing the lcd manufactures are limited to what's available, likely designed for high resolution cell phones or something.
When the 16k lcd comes out it is game over, that is a 16x20 at 540ppi which is what you get now with 8x10 at 8k and they look like contact prints. I was doing 8x10 and contact printing on AZO years ago and have a stack of prints to compare to.
BTW I also was able to drive the board with a display port and 8k converter dongle, but it was with a fairly beefy nvidia gaming card. Since intel graphics don't support custom resolutions you are stuck with nvidia or amd graphics as they come with utilities for custom resolutions..
This is why I settled on the Radeon RX 6400 as a inexpensive card for a dedicated machine that won't require a dongle and will hopefully support the higher resolutions when they come out.

 

[radiant]

The problem isn't necessarily bandwidth, it is having actual chips to drive the y axis at that refresh rate. There is no specific technical issue with making a chip but there is nothing like it and I am guessing the lcd manufactures are limited to what's available, likely designed for high resolution cell phones or something.
When the 16k lcd comes out it is game over, that is a 16x20 at 540ppi which is what you get now with 8x10 at 8k and they look like contact prints. I was doing 8x10 and contact printing on AZO years ago and have a stack of prints to compare to.
BTW I also was able to drive the board with a display port and 8k converter dongle, but it was with a fairly beefy nvidia gaming card. Since intel graphics don't support custom resolutions you are stuck with nvidia or amd graphics as they come with utilities for custom resolutions..
This is why I settled on the Radeon RX 6400 as a inexpensive card for a dedicated machine that won't require a dongle and will hopefully support the higher resolutions when they come out.

Ah, now I understand. RX 6400 was reasonable price and it packs many good things, so again I'm satisfied with the purchase. The lack of custom resolutions is of course a blocker so it is trivial what to choose. Raspberry Pi was surprise that it should work, I have to try.

I managed to enlarge the display image through the lens first time and oh man, that is sharp. Now I wish I had some digital photo that large I could use the whole surface of the screen!

There are plenty of small things I need to fix to make enlarging with this monster enjoyable, but so far so good! I'm so happy I got into this point because this was a bit scary and sketchy project at start.

 

[radiant]

The main challenge is to create a color profile and/or calibrate the thing. I'd be tempted to try the method put forth by Calvin Grier in his calibration ebook series. This seems to be an application where his approach would work exceedingly well (without the need for separations btw).

This sounds easy but it probably isn't when it comes to tiny tweaking during printing. Of course it would be interesting to try.

 

[avandesande]

If someone was looking for a turn key solution I think this would work, it has both HDMI 2.1 and USB4. I am sure there is something equivalent in the mac world as well-

Minisforum UM790 Pro

Minisforum UM790 PRO is Powered by AMD Ryzen™ 9 7940HS Processor and AMD Radeon™ 780M. Dual NVMe PCIe4.0 & Dual 40Gbps USB4. Intel® Killer™ AX1675 Wi-Fi 6E Network Card. The UM790 Pro Features An Innovative Cooling System That Actively Cools The Memory and SSD

store.minisforum.com

 

[koraks]

This sounds easy but it probably isn't when it comes to tiny tweaking during printing.

Yeah, that's right; the concept is straightforward, but the devil is likely going to be in the detail. But it's not as complicated as, say, calibrating a color gum printing process.

 

[radiant]

Yeah, that's right; the concept is straightforward, but the devil is likely going to be in the detail. But it's not as complicated as, say, calibrating a color gum printing process.

Gum print you say? I have some dichromates and gum arabic waiting for the inspiration to arise I did some gum prints on workshop but nothing at my own darkroom.

I've been looking for a FPGA project to start with a "real need". This could be it; USB based EDP1.2 driver Or maybe straight from SD card..

 

[radiant]

Raspberry Pi 4 (4GB) works fine with the display and the adapter board.

This is both money saver and space saver - you don't need a computer, graphics card or even display. Or mouse and keyboard laying around.

 

[avandesande]

It sounds like nirvana, but if you offered me a all-in-one solution for free, I would reject it immediately. When I started this project, I had a very simple setup where I would process files on my workstation and usb/wifi the files into the darkroom machine. This gets old very quick, nothing destroys your darkroom rhythm like leaving the room and sitting at a computer. Not only that, you start having to manage a boat load of files with different contrast curves or other adjustments. All of that I can access now in the darkroom in seconds. I only need a single tiff, and I can print it on anything I can think of without ever messing with it again. My darkroom is very small 8x8 and I could use the space but I wouldn't even consider it. At the very least you would need a touchscreen to apply luts or adjustments. All of the features I have added to the software has come from experience printing hundreds of prints on this new medium. I'm probably tweaking things a dozen times in the software in one session.
Also, the raspberry pi costs about as much as the graphics card, so the savings isn't that great.

 

[radiant]

It sounds like nirvana, but if you offered me a all-in-one solution for free, I would reject it immediately.

Me too. I'm not trying to go too easy and there is no fear for it. With my self built enlarger only it is very manual work alone.

But I don't want to "computerize" this either; keep it as close as film based darkroom work is.

I'm probably tweaking things a dozen times in the software in one session.

That's the beauty of this, you are not bound on the properties of negative alone, you can tweak the "negative" too.

I'm probably not going to use photoshop or any kind of "area" based manipulation (dodge/burn) on the photos I print. Only cropping and basic level adjustments only. I want to keep the film feeling and not make this like modern digital photography (spending countless hours on photoshop).

Some may say this is hypocrite but I'm a bit fed up with shooting and developing film. It is boring manual labor and film limits my way of photographing. But I love all kind of paper prints be it traditional silver gelatin or alt prints. This is like best of both worlds, really!

 

[radiant]

I printed first real papers this evening and some notes:

Sharpness - YES. It is really sharp. It looks like some really sharp film, maybe 4x5" Tmax 100 or Adox HR 50. Single pixel lines on 8x10" print are sharp and hard to see without looking closely.

Contrast - That 300:1 contrast ratio which was on some spec is probably close. And that is a bit bad news; even at Grade 0 I couldn't get straight linear gray scale to fit on paper. This means the calibration LUT will compress grayscale, meaning the print has less "resolution" in grayscale, meaning smaller amount of grayscale steps. I would guesstimate 75% so maybe 175 steps. Good news is that salt prints are going to bend in front of this display like butter!

Resolution - where can I get 8K digital photos?! I want to use all those tasty pixels on the print without loosing the DPI on the print as this is sharp as hell.

Overall feelings: this is revolutionary. Just amazing.

A huge thanks to @avandesande for letting us know about this possibility and giving very good tips for where to purchase and what to purchase. Without you I wouldn't be in this point!

 

[avandesande]

Congrats!
There are plenty of high resolution raw files online if you want something to test with, I've gotten them off dpreview (gfx50 raw) and some leica monochrome DNG on other sites.

I probably didn't explain it well but it is possible to have more than 255 levels of grayscale.
Suppose you want grayscale 100.5. You can do this by exposing 100 for 5 seconds and 101 for 5 seconds.
Using this principal the software creates a slide show of 16 different images with slightly different levels of grayscale. During exposure it cycles through each image equally. This creates a grayscale depth of 12 bit, or about 4,000 levels of grayscale. You could do more 'slides' but I think 12 bit is plenty and in testing the gfx50r only has 12 bits of real dynamic range. Lots of manufactures are supporting claiming 16 bit raw but it really boils down to what the sensor can do.

If you can please post some of your results I would love to see it!

 

[radiant]

Congrats!
There are plenty of high resolution raw files online if you want something to test with, I've gotten them off dpreview (gfx50 raw) and some leica monochrome DNG on other sites.

That is a great tip! But do I have to buy GFX50 now

I probably didn't explain it well but it is possible to have more than 255 levels of grayscale.
Suppose you want grayscale 100.5. You can do this by exposing 100 for 5 seconds and 101 for 5 seconds.
Using this principal the software creates a slide show of 16 different images with slightly different levels of grayscale. During exposure it cycles through each image equally. This creates a grayscale depth of 12 bit, or about 4,000 levels of grayscale. You could do more 'slides' but I think 12 bit is plenty and in testing the gfx50r only has 12 bits of real dynamic range. Lots of manufactures are supporting claiming 16 bit raw but it really boils down to what the sensor can do.

Yes I read about this before but haven't yet looked into this. I was thinking of it, but I couldn't figure out why I would need that as my files are 8-bit and the display is 8-bit. Now that I need to decrease the displayed dynamic range, this comes handy.

However I have to figure how to do your method in code - don't spoil yet Also my setup doesn't now support that thing as I'm just using RPi to display static image.

I was thinking should I incorporate RPi in exposuring but if I do that "slide show" exposing, I need to add more features to it. Not that big work.

If you can please post some of your results I would love to see it!

I have two prints from yesterday and have been examing them now they are dried and everything looks good. Sharp, no vignetting (phew!), plenty of contrast.

I'm going to print some more soon and I will definely post my results here. I have to make the calibration first.

 

[radiant]

Here is print of my calibration image. This is "fullscreen" so each pixel on screen is displayed. On left there is some blur but that is not on the print, some glare in the photo itself.

This print was made with about 2.5 - 3 grade so it explains why both highlights and shadows are cut, leaving the grayscale range short.

The printed photo is 21.5cm x 12cm in size.

 

[radiant]

First decent print with just straight LUT = no correction. I wanted to see how it works that way first. Shadows are of course a bit steep but at least on this photo they work fine.

Paper is 8x10" Ilford Multigrade RC V.

(the print is a bit difficult to photograph because of the glare, the paper is "pearl" so it has some shine)

 

[koraks]

That's really darn good.

 

[avandesande]

Looks good! While luts are going to be important especially in a professional setting, I discovered early on they aren't everything. A lot of the character of the print is from pushing into the toes of the paper just like you would with film and luts prevent that. You have me thinking even about a lutless approach, where you digitize your negative, fix levels and bring out the contrast using darkroom techniques.
As far as 8 bit vs higher bits, I think 175 steps is not the right way to look at it. If you look at the slope of the papers response curves, especially in the all-important midtone region, you are getting far less, like under 100. And if you want to take a undoctored low contrast image and use higher contrast filters it will be even worse.

 

[radiant]

That's really darn good.

Thank you, sir. I really appreciate that.

Looks good! While luts are going to be important especially in a professional setting, I discovered early on they aren't everything. A lot of the character of the print is from pushing into the toes of the paper just like you would with film and luts prevent that. You have me thinking even about a lutless approach, where you digitize your negative, fix levels and bring out the contrast using darkroom techniques.
As far as 8 bit vs higher bits, I think 175 steps is not the right way to look at it. If you look at the slope of the papers response curves, especially in the all-important midtone region, you are getting far less, like under 100. And if you want to take a undoctored low contrast image and use higher contrast filters it will be even worse.

LUTless approach isn't bad at all for sure. In digital enlarging the traditional darkroom work is easier overall as the negatives are well "exposed and developed". There might be small variances, but with digital you have perfect negative every time.

I'm controlling digital exposure with my old trusty self built f-stop / contrast control box. I can easily change exposure in stops and the calculation takes account contrast too (I set contrast with value 0-100). By using that with digital negative it is almost perfect for general use. Close enough for sure.

What I've been missing in traditional darkroom work is broader way to manipulate negative. If you screw the shot or development, that's it. You can only do some adjustments but very often I've ended up with just giving up on the negative. Of course this is 100% self caused but I'm just not that patient what it requires and I cannot help myself.

Digital enlarging combines both worlds; traditional darkroom and the power of digital photography.

What would be interesting is to do a subjective test for prints with different amount of gray levels. I think I remember reading somewhere sometime that human can differentiate 60-70 shades of gray in photo and best from highlights. 100 shades of gray starts to be a bit too low, but is it really? Interesting test that would be.

All is not lost ofcourse as one can use that your @avandesande ingenious shade-increasing technique so this is not a huge problem really.

 

[avandesande]

I am sure 8 bit is fine for normal exposure on grade 2 but I started seeing problem when I was using harman direct positive paper which is very high contrast. It also shows up when lith printing, especially with old developer.
The math is simple 16bitpixelval + (stepnumber * (255/numberofsteps))

 

[radiant]

I am sure 8 bit is fine for normal exposure on grade 2 but I started seeing problem when I was using harman direct positive paper which is very high contrast. It also shows up when lith printing, especially with old developer.
The math is simple 16bitpixelval + (stepnumber * (255/numberofsteps))

Yeah, direct positive paper. I wouldn't even think about it on this. I've shot some DP in-camera and that paper is one wild horse

Do you mean low gray scale shows up in lith? Interesting, how? I'm going to do liths as well, that is favorite process.

 

[radiant]

I started working on LUT / calibration. Here is my 16 step digital print as graph:

Y-axis is measurement. Two lines = the other one is just 0.4 stops more exposure - measurement seems reliable.

What I can see from this photo is that the response of paper "dips". If I print with linear gray scale match (=no calibration), the print will become darker than what I see on digital file. Which matches my experience.

 

[avandesande]

It might not necessarily be all the paper. Keep in mind these lcd are used for only 1 bit application in 3d resin printing, not sure how much testing they have done for continuous tone.. I've always wanted to measure the lcd directly for linearity but haven't had the time for it. Of course any problems with lcd can be fixed with luts or software.

 

[radiant]

It might not necessarily be all the paper. Keep in mind these lcd are used for only 1 bit application in 3d resin printing, not sure how much testing they have done for continuous tone.. I've always wanted to measure the lcd directly for linearity but haven't had the time for it. Of course any problems with lcd can be fixed with luts or software.

I assume it is pretty linear now I have seen how it prints out. I have light meter which I made for printing use so I could measure the linearity.

However by my experience the linearity is very good anycase. Easy to calibrate (if even needed).