DIY 31 Megapixel Enlarger

[radiant]

So I measured my calibration, implemented it and exposed new prints. It was easily worth it. The result of it is very easy to see straight away.

One thought we have been discussing here few posts ago got even stronger feeling: this is wet darkroom printing after all. You need same experience as working with film; skill to analyze and adjust. Altough digital negative enables using digital images (really?!) and skip the work with negatives (exposing/developing which is completely its own skill to master) - you are still at darkroom, with very good negative, but still printing on enlarger on whatever medium you choose.

BTW: Have you ever thought you have some dirt fallen on your print and just ignored it - and then finding out IT IS YOUR MOUSE CURSOR I found the cursor with grain focuser as I was trying to check my focus. It is very difficult to focus btw. The resolution is just amazing on 8x10" print - I cannot see pixels even with grain focuser.

(spoiler: LUT print on left)

 

[radiant]

I made new prints today. They all look very decent but I noticed my deep shadows are loosing details. It needs a just a tiny hint of contrast. Same time highlights are just a little too bright, for example clouds and bright sky loose contrast.

Only idea I came out is using S-shaped curve on levels. How much? That requires experienting = a big trashcan.

My current "digital negative development proces" is also a bit slow for making adjustments in the darkroom, maybe.

My general plan is to print photos that do not need full tonality or such - but as this is so good it is tempting to nail some prints so that they look gorgeus. And a bit too gorgeus to be wet prints. I realized that part of the charm of wet darkroom prints come out from the fact that they aren't perfect. With digital you can get so perfect prints that they "escape" totally from the looks of "normal" darkroom prints - and it is so noticeable that people are probably asking "are these inkjets?"

 

[avandesande]

I used the QTR step wedge script, photoshop and a 21 step wedge (included with the tool)

QuadToneRIP Downloads

www.quadtonerip.com

Yes, it gets very tricky near the toes.
Below is curve for ilford warmtone and output chart. Don't worry about the image it is something random without level adjustment.
I hope to start building a tool that creates a lut based off of a gradient that you would crop to size and the software would read automatically.

 

[radiant]

I used the QTR step wedge script, photoshop and a 21 step wedge (included with the tool)

QuadToneRIP Downloads

www.quadtonerip.com

Yes, it gets very tricky near the toes.
Below is curve for ilford warmtone and output chart. Don't worry about the image it is something random without level adjustment.
I hope to start building a tool that creates a lut based off of a gradient that you would crop to size and the software would read automatically.

That curve is pretty familiar to anyone who has been measuring around in darkroom That "lab" curve seems quite linear, is it measured patch values in grayscale?

I'm using my crappy processing.org code tool that I can make measurements very fast from image to a text file. It makes measuring multiple strips a breeze. It doesn't have any automatic system but it only requires just few manual operations which are fast to do. I never had need to do it completely automatically - it would be cool of course but implementing that takes time of maybe 10.000 "semi-manual" measurements

 

[P1505]

This is amazing! Wow. Tempted to have a go myself. If I could run the screen from a Raspberry Pi I could chuck some software together.

You could invert the image (to a positive) to compose then flip. Wow. Now to find a screen, and enlarger to use

This is really inspiring thanks for sharing.

 

[radiant]

This is amazing! Wow. Tempted to have a go myself. If I could run the screen from a Raspberry Pi I could chuck some software together.

You could invert the image (to a positive) to compose then flip. Wow. Now to find a screen, and enlarger to use

This is really inspiring thanks for sharing.

Yes, it works on Raspberry Pi 4.

For now I've always composed the photo on computer and enlarged in full screen. That way my exposure time is constant and basically done. It is like a copy machine, almost Of course you probably want to do reprints after adjustments but the cool thing about this is that the negative is always in same dynamic range + basic light level.

 

[P1505]

So all I need is the screen, Raspberry Pi, driver board and cables and a big enlarger? Or a copy stand and I build and enlarger head. I guess I just need to focus the light.

Hmm. Where to find an enlarger.

 

[radiant]

So all I need is the screen, Raspberry Pi, driver board and cables and a big enlarger? Or a copy stand and I build and enlarger head. I guess I just need to focus the light.

Hmm. Where to find an enlarger.

That's right. I even built myself an enlarger for that from wood. Works fine, it is even decently adjustable. My setup:

Light source: led strips in a light box
Lens: Large format lens, I'm using Nikkor 210mm which covers the whole screen without vignetting. Hint: Sinar lenses are cheap because you don't need shutter and aperture can be set fixed.

You need to make python script or other to "expand" your image; The RGB values contain each one true pixel as in display pixels, so you need to "pack" those into the RGB.

 

[P1505]

That's right. I even built myself an enlarger for that from wood. Works fine, it is even decently adjustable. My setup:

Light source: led strips in a light box
Lens: Large format lens, I'm using Nikkor 210mm which covers the whole screen without vignetting. Hint: Sinar lenses are cheap because you don't need shutter and aperture can be set fixed.

You need to make python script or other to "expand" your image; The RGB values contain each one true pixel as in display pixels, so you need to "pack" those into the RGB.

Can you expand upon the last bit please?

Do you mean combine the three channels into one, taking RGB and creating one greyscale pixel?

 

[radiant]

Can you expand upon the last bit please?

Do you mean combine the three channels into one, taking RGB and creating one greyscale pixel?

The display isn't 8K display for your computer. But still it is. The "width" in pixels is three times smaller than what it is in reality (2560 x 4320).

The trick is that you computer transmits RGB image to the display board. The display board then expands the shown pixels from that. The trick is that in RGB colorspace the R is actually the "first" grayscale pixel, the G "second" etc. So grayscale is packed into RGB values.

See this post and the image https://www.photrio.com/forum/threads/diy-31-megapixel-enlarger.197305/page-2#post-2715110

 

[P1505]

Ahh ok. So the first pixel’s red value is the 255 value of the first pixel of the image. The first pixels green value is the 255 value of the second pixel of the image etc. Ok, it will be fun to make a dynamic unwrapping function to cope with different sized images and handle interpolation. Hmm. Yeah this’ll be fun. I like the idea of that 12k screen also.

 

[radiant]

Ahh ok. So the first pixel’s red value is the 255 value of the first pixel of the image. The first pixels green value is the 255 value of the second pixel of the image etc. Ok, it will be fun to make a dynamic unwrapping function to cope with different sized images and handle interpolation. Hmm. Yeah this’ll be fun. I like the idea of that 12k screen also.

Yup, you got it right.

chat.openai.com + python is probably winner solution

 

[avandesande]

Here is a thread talking a bit about Raspberry PI setup, they mention the python script. It might be worth picking the image apart there might be some other goodies in there as well.

Raspberry Pi 4K(3840X2400) "HDMI_TIMINGS" in config.txt - Raspberry Pi Forums

forums.raspberrypi.com

 

[michaelbirkmose]

@avandesande and @radiant - I came across this thread today, and I am very interested in this!
Been mailing with the manufacturer (Sumaopai) about their 6.8" 9K LCD today, and it should apparently be possible to drive it with a rasberry pi 4b. I am seriously considering ordering one and getting started with the playing Fun fact the manufacturer was familiar with this thread!

Before I go down the rabbit hole I have a few questions @avandesande if you don't mind?
* I noticed you do some tricks with multiple shifted exposures to get a 12bit output. How has your experience been with this process in terms of tonality? My interest is in portraits, so smooth gradients are much appreciated.
* You mentioned initially you burned out the driver with an intel card? I wonder how that happened - is it not simply an hdmi port?
* How fragile is the display itself? As with anything darkroom dust is going to be a factor - does it handle well, is it easy to keep clean, etc?
* I saw you mounted the lcd panel in 1/8" polycarbonate sheet. Is it mounted on top of the sheet, or did you make a cutout to fit it? How about scratches/dust if its mounted on top?
* You mentioned: "The first issue and most pressing is a artifact of the lcd screen manufacture. It is actually two grids sandwiched together between glass". Does mean its two panels stitched together? How did you counteract this, and can it completely be counteracted or do you still see artifacts in the print?

Thanks for all the information you have shared so far.

 

[avandesande]

* I noticed you do some tricks with multiple shifted exposures to get a 12bit output. How has your experience been with this process in terms of tonality? My interest is in portraits, so smooth gradients are much appreciated.

Tonality is excellent, 8 bit is fine if you are printing on normal contrast papers.

* You mentioned initially you burned out the driver with an intel card? I wonder how that happened - is it not simply an hdmi port?

There is some kind of handshake/negotiation that goes on between the lcd driver and the video port. In this case it failed and drove the lcd driver at the wrong frequency and it overheated. I haven't had this problem with nvidia/radeon cards. Keep in mind I haven't tested all the different lcd drivers and hdmi ports etc. I just know what combinations works.

* How fragile is the display itself? As with anything darkroom dust is going to be a factor - does it handle well, is it easy to keep clean, etc?

The lcd panel is extremely thin and fragile. I haven't broken one yet. If you are careful with it it's not really a problem. Since I don't use the enlarger for anything else it just stays in there... I've cleaned it once in the last year. It only takes a couple minutes to remove, just like a traditional negative. The LCD costs about as much as a piece of AR glass the same size (maybe less). If I break it I will just get another.

* I saw you mounted the lcd panel in 1/8" polycarbonate sheet. Is it mounted on top of the sheet, or did you make a cutout to fit it? How about scratches/dust if its mounted on top?

The main panel is cut out slightly smaller than the lcd and it sits on top. It is centered in place by more pieces of polycarbonate glued in place with acrylic adhesive. The main panel was cut out with a hand jigsaw, nothing fancy. You can use a traditional hand planer to clean up the edges of the polycarbonate to make it perfectly smooth.

* You mentioned: "The first issue and most pressing is a artifact of the lcd screen manufacture. It is actually two grids sandwiched together between glass". Does mean its two panels stitched together? How did you counteract this, and can it completely be counteracted or do you still see artifacts in the print?

I added some correction to the software and it's really not that big of an issue (you can see sometimes if you look very hard for it under very specific conditions). From what I can tell, the 6.8" 9K LCD is just a single grid. I ordered one a couple months ago but it has been stuck in customs so I can't verify results yet with this LCD.

 

[michaelbirkmose]

Thanks for the feedback @avandesande - I think Ill just go ahead and order the driver and the lcd and see where the journey takes me

From what I can tell, the 6.8" 9K LCD is just a single grid. I ordered one a couple months ago but it has been stuck in customs so I can't verify results yet with this LCD.

How did you see this?

 

[avandesande]

Thanks for the feedback @avandesande - I think Ill just go ahead and order the driver and the lcd and see where the journey takes me

From what I can tell, the 6.8" 9K LCD is just a single grid. I ordered one a couple months ago but it has been stuck in customs so I can't verify results yet with this LCD.

How did you see this?

You can see in the 10.1 8k there are two data busses driving the grids. The 6.8 9k only has one.

 

[michaelbirkmose]

Right! I was wondering - how about pixels - do you observe those in the print? Looking at this guy's experiments, it appears like pixels are visible (different workflow though): https://hackaday.io/project/171069-digital-picture-to-analog-darkroom-print

 

[avandesande]

It looks like he was contact printing, any lines are much less apparent in an enlarger. Also this is a older generation lcd, I don't know how large the grid wires are and due to it's size they probably carry more current. I've printed 16x20 and you can see some artifacts if you stick your nose in the print. 8 x10 is nearly 600dpi with 8k lcd you are hard pressed to see anything in this size print.

 

[radiant]

Right! I was wondering - how about pixels - do you observe those in the print? Looking at this guy's experiments, it appears like pixels are visible (different workflow though): https://hackaday.io/project/171069-digital-picture-to-analog-darkroom-print

On my 8x10" enlargements with the 10.1" screen there is no way to see the pixels. I've even had mouse cursor on enlargement and it is reeeally difficult to find from the print, if even possible. That gives some idea of how sharp the result is.

 

[radiant]

Fun fact the manufacturer was familiar with this thread!

Yeah I think he knows all of the users There aren't many. Back when I asked him questions he said "one other user is doing the same" - and I'm pretty sure he was referring to avandesande.

 

[avandesande]

I've been thinking about an approach for large prints where a flashed/developed film is epoxied to the LCD (probably bottom). You would focus on the grain in the film instead of the grid lines. This would allow large prints to 'break up' naturally.

 

[radiant]

I've been thinking about an approach for large prints where a flashed/developed film is epoxied to the LCD (probably bottom). You would focus on the grain in the film instead of the grid lines. This would allow large prints to 'break up' naturally.

Do you mean expose the film so that grain is left on the film? Like mid-gray? Focusing on the grain would then cause a the digital one to be a bit out of focus and then masked with sharp grain?

 

[avandesande]

Do you mean expose the film so that grain is left on the film? Like mid-gray? Focusing on the grain would then cause a the digital one to be a bit out of focus and then masked with sharp grain?

That's the idea, I would try it with tape first though

 

[radiant]

That's the idea, I would try it with tape first though

That is pretty interesting idea, a bit crazy. I have some 5x7" sheets I could try myself too. Just need to expose one with correct density. Or maybe do few to see what works the best.

Downside is that it decreases the light amount in the exposure for few stops perhaps.