100+ megapixels from 6x9 film

This picture is the first in a set here:

www.flickr.com/photos/119759627@N06/sets/72157642579675275/

...that is a pictorial reply to a question by flikr user miloniro about how I digitzed this photo:

www.flickr.com/photos/119759627@N06/13218284835/

 

It shows how I got a very clean hi-res result by digitizing my slides at home using an old enlarger machine lens on my camera.

 

This technique was inspired by this excellent article by Gianluca Bevacqua:

petapixel.com/2012/12/23/why-you-should-digitize-your-fil...

 

* * * * * * * * * * * * * * * * * * * * * * * * * * * *

In this photo, you can see the basic requirements:

 

1. A very sharp color reversal slide. To try and get a sharp slide, I was careful to use:

a) Medium format - I'm using 6x9 format

b) Fine-grained film - I'm using Velvia

c) Sharp lens - I'm using a Fuji GSW690III

d) Sharp apeture - f/16 is good on this lens

e) Solid tripod - Manfrotto 290 series

f) Cable release

G) Top-notch processing lab - Vision Image Lab, Redfern, Sydney.

 

2. Tripod with removable stem that can turn upside-down as pictured. Props to these guys for enlightening me to the rather obvious:

forum.xitek.com/thread-620804-1-1-1.html

 

3. 4-way macro focusing rail - vertical is for focusing, horizontal is to move side-to-side to get the next frame. That way I can push the lightbox forward and backwards (up/down in slide orientation) only when I have the edge of the frame in view, so I don't get lost.

 

4. Fuji X-E1 camera - I chose it specifically for this because the X-Trans sensor doesn't have an anti-aliasing filter, so it's sharper than most. I guess a D800e or A7r would do nicely as well :)

 

5. Some macro extender rings

 

6. Schneider Componon 50mm enlarger lens

 

7. A piece of heavy cardboard with a hole 2mm longer and wider than my slide, to hold it flat and to mask the rest of the lightbox to prevent lens flare

 

8. Lightbox - Kaiser LED Slimline

 

9. A heavy screwdriver set to act as a moving platform for the lightbox... the lightbox is really light and easy to knock out of position but the screwdriver set is heavy and slides smoothly on the carpet, making it easy to make small vertical moves when I get to the edge of the slide after each pass.

 

10. USB cable release (I also set the camera to 10 second self timer - it takes quite a while for vibration to subside after moving the camera side to side - I would love to have a rig where I could wind the lightbox around instead... I reckon this guy is on the right track: www.photomacrography.net/forum/viewtopic.php?t=15921)

 

11. LARGE "ROCKET" DUST BLOWER - the most critical piece of equipment - seriously!!

 

12. Ruler I can photograph to see what field of view my enlarger rings are giving me (I have a set of three: 7mm, 14mm, and 28mm, and can combine them in different combinations)

 

13. 2-way spirit level to check that I have the camera straight

 

14. a couple of thin pieces of black card to mask any bright out-of-shot parts of the slide that are causing lens flare

 

15. An extra quick-release plate and mount (hidden under camera in this photo)

 

16. Fuji X-mount to Leica M39 adapter ring (enlarger lens has M39 thread)

 

* * * * * * * * * * * * * * * * * * * * * * * * * * * *

TECHNIQUE:

 

A) Unlatch and splay tripod legs, pop end cap of tripod stem, remove stem, invert, replace stem, replace cap.

 

B) Join some enlarger rings together (for 29mm FOV I found 14mm + 28mm worked), fit the M39 adapter to them, and screw enlarger lens into M39 adapter. I use the blower to remove any dust off top of lens and out from inside of ring/lens assembly.

 

C) I attach the rings/lens unit to my X-E1 camera. I then set the lens apeture to 5.6, which I have found to be the sharpest for this particular lens(?)

 

D) I fit the tripod mounting plate to underside of focusing rails. I fix a spare quick release bracket to the top of focusing rails (this is a right pain in the neck, which is why I use a quick release plate instead of fitting the camera directly). I then fit the spare quick release plate to camera.

 

E) I adjust the tripod head into the position shown, so that I will have pitch and yaw control - there is another way it will go, which doesn't work... Fit focusing rail assembly to tripod head. Fit camera to focusing rails. Adjust vertical rail so it's in the middle of it's range, and the horizontal rail so it's at the end of it's range.

 

F) I put a kleenex on the camera LCD to protect it, then put the spirit level flat on the LCD screen and adjust the tripod head until it shows the camera is level. With all the torque on the tripod head at this point, some swearing is usually involved...

 

G) I attach the USB release cable. I set the camera to manual focus and turn it on. I put the lightbox under the camera and turn it on too, and take a "custom white balance" setting from it so the colors will turn out ok. I set the camera to 10 second self timer, disable auto power off, disable the EVF eye sensor, set quality to Fine+Raw 3:2, set film emulation to standard, enable "shoot without lens", and set shutter speed to auto. I have started playing around with the noise reduction setting too lately - I am still trying to work out what the best setting for that is. I haven't played with sharpness, shadow etc. yet.

 

H) Now it's time to get out the slides: I first move the lightbox platform into place, then blast it with the dust blower, wipe it with a lens cloth, then blow it again (yes, really - at 130 megapixels, dust is a nightmare). Then I carefully remove a set of slides from it's sleeve and carefully blow the dust from both sides (yes, it's there...).

Then I place slides centrally on the lightbox, I place the slide right way up (matte side down to fight newton rings), and aligned with the camera so I don't need to rotate images later.

Then I pick up the cardboard mask and blow dust off both sides of it, then carefully place it to mask around the slide I want to digitize.

 

I) Focus time: This enlarger lens has this cool lever on the side that pops the apeture wide open when you turn it one way, and closes it back to your pre-set apeture when you turn it back. The idea is to pop the apeture wide open when focusing so that the shallow depth of field will highlight your mistakes, then to close it back before shooting so that it hides your mistakes. So I pop the apeture wide open, then I turn the lightbox back on, and move it so that a detailed part of the slide is under the lens. Then I adjust the tripod stem up and down very slowly (without driving the lens lever through my slide) until I see the live-view image on the LCD come into focus, and lock off the tripod stem in that position.

 

J) Focus part deux: Then I press the dial on the back of the camera to get focus assist (the image is enlarged), and unlock the vertical focus rail (while holding it with my other hand so it doesn't drop the camera through the lightbox), and use the dial to move the camera up and down by fractions of a mm until the focus peaking is looking pretty good. Then I lock the rail again, and make even smaller adjustments against the resistance of the lock (by using lots of force) until I am happy with it. Then I cancel focus assist.

 

K) Exposure: I pick a mid range part of the slide and move it under the lens, and see what shutter speed the camera thinks it needs. I then manually lock shutter speed to the closest approximation the shutter speed dial provides, so that exposure won't vary across the different frames I take.

 

L) Scanning: I move the lightbox platform until one corner of the slide is neatly and squarely lined up in the corner of the LCD image. I turn down the room lights to prevent reflections, use the black card strips to mask any out-of-view areas of the slide that are causing lens flare, and (while holding down card strips) gently blow of all the dust that has settled on the slide while I was focusing (yes, it's there...) Then I use the cable release to take the first frame (finally!!)

 

M) Scanning II: I check that the horizontal focus rail is unlocked. Then I pick a feature that is in the middle of the LCD view, and move the dial on the focus rail until that feature is shown touching the edge of the LCD frame. Then I blow dust off again, then take another shot. Usually the vibration from twisting the focus rail dial is so bad that I have to let it calm down before taking my shot, even with the 10 second timer. I use the time for dusting :)

 

N) Scanning III: I repeat the step above till I hit the far edge. Once I have taken the last shot of that row, I carefully move the slide vertically to get ready for the next row - again I pick a feature in the middle of the LCD image and move the lightbox until that feature hits the top.

Every time I have tried to cheat by using less than a full 50% overlap between frames, I have gotten horrible tearing in the final stitched image and had to throw out the whole project and start over - but YMMV...

I always include some of the masked border in the frame so that I have a point of reference when moving the lightbox - it helps me keep everything square. Once the lightbox is in position for the next row, I take the frame and then go back to moving sideways with the focus rail till I hit the far edge - and so on and so on and so on and... it takes about 20 minutes to get around 30 frames, what with waiting for the image in the LCD to stop shaking and with dusting and so on.

 

O) I found if I forget to dust each frame I will regret it.... Likewise if I overdo it and blow the cardboard and slide onto the dusty carpet :-/

 

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

 

For more pictures of my set-up, go to the set here:

www.flickr.com/photos/119759627@N06/sets/72157642579675275/

 

* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *

STITCHING THE FINAL IMAGE

 

I combine all the individual frames into a single image using a panorama stitching program. I use an open source (i.e. high-quality, zero cost) software package called "Hugin", which you can get for Windows, Mac, or Linux here:

hugin.sourceforge.net/download/

I also had to do some things to make sure my graphics card drivers were working (I use linux - it's worth it...) but that is another story altogether - suffice to say that Hugin crashing is probably graphics card driver problems - but that is not Hugin's fault.

 

Panorama programs expect that all your pictures were taken at the same spot, but the camera has pivoted (pitch, yaw and roll) between frames. In my case, it's exactly the opposite - I have no pitch or yaw, but I move x and y location with every shot. I also have a bit of roll, but I don't have any z location (depth) movement which is another parameter that panorama programs try and force on you.

 

So basically, whether I am using Hugin or some other stitcher, I now need to tell the program that x, y, and roll adjustments are ok, and that pitch, yaw, and z adjustments are not. And I need to tell it that I had a "rectilinear" lens that was very very long (so it doesn't correct for barrel distortion), and that I want my final image produced by a "rectilinear projection". I also want it to calculate "control points" automatically - I have been getting over 2,000 of them for most projects, I don't want to do that by hand!!

 

In case it helps anyone, the workflow I use for Hugin is as follows:

 

1. Switch UI mode to "expert"

 

2. Click the "Fast panorama preview" button to open a preview window. I put this on my other monitor (I find having a 2nd monitor is indispensable, and they are so cheap these days!!). I don't worry about what I see in this window until right at the end, it goes a bit nuts as I change some of the settings below.

 

3. Click "load images", and select all the JPG files I put into a single folder (one day soon I am going to get around to exporting the RAW files to 16bit TIFF first and using those instead, I swear...)

 

4. Tell Hugin that I used a ridiculously (actually, effectively infinitely) flat lens; I set the focal length to 1000 and the multiplier to 1.

 

5. Tell Hugin that every picture was taken with a different camera, so it doesn't try any stupid interpolation tricks and stuff everything up; this means right-clicking every line in the image listing (except the first one) and then selecting lens->new lens

 

6. At the bottom of the screen are two drop-down lists for "optimize". I click the "Geometric" one and choose "custom parameters" so that I can specify no yaw, pitch, or z. This displays a new "Optimizer" tab at the top of the page.

 

7. I go to the Optimizer tab. This tab shows yaw, pitch, roll, x, y and z for each image. All values are initially zero - but the bold ones with lines under them are ones that are "active" - Hugin is free to change them.

The fields can be toggled from active to inactive by holding down the Control key and left-clicking the field. The top row (first image) I set all fields to inactive. For all other images (rows) I de-activate yaw and pitch, and activate x and y, and leave z as inactive. This involves hundreds of clicks and is a real pain.

 

8. I go back to the "Photos" tab, and I click the "create control points" button. After a while it comes back and tells me I have hundreds or thousands of them. It also updates the list of images to show how many control points each image has - if any are zero, I am in trouble. That only happened to me once fortunately, because I don't know how to fix it!

 

9. Next to the "Optimise - Geometric" drop-down list, there is a "Calculate" button. I click this next, and eventually Hugin comes back with some numbers and asks if I accept them. I click yes, and at this point the preview window on my other monitor shows a lot of text that tells me the panorama has a "Very Good Fit".

 

10. I go to the "Stitcher" tab, and change the "Projection" drop-down list to "Rectilinear".

 

11. I double-check that "Projection" is now showing as "Rectilinear". If it's anything else, then the stitching results will be spectacular. Confusing, ugly, and undesirable yes, but quite spectacularly so!!

 

12. I click "calculate field of view". After some time, the numbers change to single digits and I can actually see a panorama in the preview window instead of just a dot or tiny thumbnail.

 

13. I click "Calculate optimal size". After some time, the numbers change to somewhere between 1 and 3 times the values I was expecting, and the preview contains extra black space accordingly.

 

14. I click "Fit crop to images". After a VERY long time, a white crop border is drawn in the preview window right where it should be to produce a nice panorama.

 

15. I make sure only the first "Panorama outputs" checkbox is checked (Exposure corrected, low dynamic range), and that TIFF and LZW are set for format and compression respectively.

 

16. I click the "Stitch!" button. Hugin opens two new windows, one for the background job monitor (the stitching runs as a separate background job) and another that shows console-like output of all the individual stitching commands that are being run. Stitching commences.

 

17. I realize I have 20 other programs open and will probably run out of memory, and franticly close all the un-needed ones, including my internet browser because all those tabs are a real memory hog! Then I go make a cup of tea.

 

18. After 10-20min (depending - my machine is an Intel i7 3rd gen quad core with 16mb ram and a solid-state drive... and I make sure that all files are on the SSD not the HDD, it makes a huge difference...) the rendering should complete. Then I open the resulting TIFF file in my photo-editing program (again I am using open source - Gimp, it's a solid photoshop alternative if you aren't doing CMYK) so that I can correct any tiny bit of rotation (usually 0.2 degrees or so) and crop off the black borders, plus do any color corrections.

 

19. Then I export it as a JPG at max quality and 4/4/4 chroma, and in the rare case that I think it could stand public scrutiny, I upload to Flikr, and I am done!

 

 

Done