Advice for beginner photographing small artwork

[617]

Good morning. I was wondering if anyone could offer some guidance for a project I am working on. Basically, I am making small (max 4"x5", min 2x2.5") painting/collage/sculptures. These objects are mostly flat, with features within a 1/2" (12mm) depth of field.

Photographing these guys would not be challenging if I simply took a single image of each. I downloaded helicon focus/remote, and with the focus peaking I can easily find focus and make a stack of images. Helicon even works with raw files, which is great, since I am using a color checker to keep color as accurate as I can.

The problem is that I am trying to print these images quite large at 300dpi, and part of the artistic intent here is to render very high levels of texture and detail at all parts of the picture. As a result, I am planning on photographing these objects in a 3x3 or 2x2 or 1x2 grid with my camera/lens (Canon 70D, Ef-s 60mm macro). This combo can make a good looking 12x18 but I want a spectacular 20-30" image I need to step it up. Photographing in a grid allows me to get arbitrarily high quality; I think a 3x3 grid will give me as much resolution as I could want.


I have a few questions I could use help with - macro is a new world for me.

1. Does anyone have experience combining panoramic techniques with focus stacking? Doing one or the other is easy, but doing both is tricky. Should I make the panoramas first, then stack? Or focus stack and then combine in panormas?

2. Would a longer lens ameliorate potential problems with perspective changes as I move the art around under the lens?

3. Should I just plan on investing in a higher quality camera/lens? If so, which? Double the linear resolution of the 70D would presumably be a 80mp camera, but perhaps going to full frame with a good lens would do the trick? I could probably swing a used d800 or something like that.

4. From the measurements I've seen, sony bodies such at the A7 series seem to offer the highest resolution per dollar, but they are not supported by helicon focus.

5. What are the best bargains in canon/nikon macro lenses? Does helicon focus and other software have difficulty stepping sigma/tamron lenses?

[aveslux]

Stitching and stacking are quite common techniques these are the search terms you want to use to find relevant discussion.

At the high commercial end there's stuff like http://www.gigamacro.com/gigapixel-macr ... ng-system/

But the basics are the same for your own setup, you move either the subject or the camera depending on how easy it is. generally moving the camera is preferred for the lighting consistency.

There are 2 ways to achieve stacking internal camera focus stepping and rail based sliders that move the camera lens system. You might find rail based stacking gets better results than camera focus stepping.

Your lighting will important, ideally you'd would want some diffused strobes to ensure consistency for both stacking and stitching.

You actual image resolution is based on your angular resolution and resolving power more megapixels doesn't always mean more details.

[617]

Aveslux, thank you for your reply. You've given me a lot to think about.

Regarding cameras, I think that a panoramic image is still the way to go for high quality. Hard to compare directly, but it appears that a camera and lens which does better than a 3x3 or even 2x2 panorama with my current set up is serious cash.

That gantry controlled macro set up is really interesting. I'd like to avoid having to move my camera in XYZ but that seems to be the best way to keep everything consistent. Right now it looks like to enlarge a small piece of art I would need something around 3" of x travel, maybe a bit less Y, and Z travel around the depth of field I'm trying to capture, maybe 1".

I will see how consistent my lighting is as I move the art beneath my stationary camera. If it works, it works - I'm using two video lights, both bigger than the art itself, but if shadows become screwy as the art moves I'll have to engineer something better.

[apt403]

I have a few questions I could use help with - macro is a new world for me.

1. Does anyone have experience combining panoramic techniques with focus stacking? Doing one or the other is easy, but doing both is tricky. Should I make the panoramas first, then stack? Or focus stack and then combine in panormas?

2. Would a longer lens ameliorate potential problems with perspective changes as I move the art around under the lens?

3. Should I just plan on investing in a higher quality camera/lens? If so, which? Double the linear resolution of the 70D would presumably be a 80mp camera, but perhaps going to full frame with a good lens would do the trick? I could probably swing a used d800 or something like that.

4. From the measurements I've seen, sony bodies such at the A7 series seem to offer the highest resolution per dollar, but they are not supported by helicon focus.

5. What are the best bargains in canon/nikon macro lenses? Does helicon focus and other software have difficulty stepping sigma/tamron lenses?

1.) Quite common! Stacking then stitching is generally the way to go. PTGui and Microsoft ICE (Image Composite Editor) seem to be the most popular options for the stitching.

2.) I'm not quite sure how focal length effects perspective change, someone more knowledgeable will surely chime in there. For 1:1 and higher stitched panos, parallax error is one of your biggest enemies however. The two most common ways of dealing with it are by either rotating the camera around the entrance pupil of the lens, or using telecentric optics. Here are a few good threads that helped me out quite a bit. First link is one part of a three thread series.

http://www.photomacrography.net/forum/v ... php?t=1472
https://www.photomacrography.net/forum/ ... hp?t=37460

At higher magnifications, the rotation around the entrance pupil method becomes challenging (imo), and telecentric optics become much easier to work with.

3.) You mentioned 20-30" prints at 300 DPI - That's about 6000 and 9000 pixels, easy enough to accomplish with your 70D. For the minimum subject size you mentioned (2x2.5") that's a reproduction rate of about ~1:2.7, so your 60mm lens should be more than adequate for the job. My personal goal is making 6ft+ prints with a 24mp D5600, and I don't find the resolution all that limiting. The one benefit of a higher resolution camera is that fewer tiles are needed for the same print size, with the caveat that eventually the camera sensor will out resolve the lens. I haven't run into this problem at 1:1 or lower.

5.) From a purely price:performance perspective, old enlarger lens are hard to beat, such as the EL-Nikkor f/2.8. Maybe $30-$50, used on eBay. Johan and Mark Goodman have put together some good info about them at their sites, extreme-macro.co.uk and coinimaging.com, respectively. They're manual focus however, requiring a rail system of some variety.

Regarding the lighting - As aveslux has said, diffusion is key, especially if you're going to be moving the subject rather than the camera.

If I were to do this, starting from the ground up, I'd attempt to add a limiting aperture at the focal plane of that 60mm macro lens to make it telecentric, and invest in a WeMacro rail or something similar, mounted vertically. With heavily diffused, even illumination, you should be able to translate the subject in X-Y by hand. You do lose the 'closer is bigger' perspective using telecentric optics; there is one thread on here where a fellow documented his rotation around the entrance pupil setup (essentially a multirow panorama rig like you'd see for landscape, etc), but I can't immediately find it.

[Lou Jost]

apt403 pointed out that you don't need too much additional resolution to meet your goal. If that is the case, then you can get the extra resolution without stitching by using a camera that does pixel shifting.

[617]

At higher magnifications, the rotation around the entrance pupil method becomes challenging (imo), and telecentric optics become much easier to work with.

If I were to do this, starting from the ground up, I'd attempt to add a limiting aperture at the focal plane of that 60mm macro lens to make it telecentric, and invest in a WeMacro rail or something similar, mounted vertically. With heavily diffused, even illumination, you should be able to translate the subject in X-Y by hand. You do lose the 'closer is bigger' perspective using telecentric optics; there is one thread on here where a fellow documented his rotation around the entrance pupil setup (essentially a multirow panorama rig like you'd see for landscape, etc), but I can't immediately find it.

Apt403, thanks for your input. This is very helpful. I hadn't considered that the perspective changes coming from XY translation are identical to those introduced in panorama photography. A gimbal head aligned to the entrance pupil of the lens might be a very practical solution, although it might be a bit big for the set up I want.

Regarding telecentricity - is there a guide you know of showing how to convert a normal telephoto lens to one with a 0 degree field of view? This might be a really great effect, lending a sort of scalelessness to the images, although if I'm not mistaken, it will mean I'm shooting everything at 1:1 essentially? It's hard to imagine what that even would look like.

You said I could add a 'limiting aperture'; I take it this mean introducing another optical element between the lens and the camera? And somehow this gives a telecentric projection?

I'm reading through all these threads now, this stuff is fascinating.

[RDolz]

Hello, about the telecentricity and about how to convert a conventional lens into telecentric, here are several links that can help you:

To convert a lens into telecentric, I followed the excellent descriptions provided by Rik in:

http://www.photomacrography.net/forum/v ... php?t=1032
http://www.photomacrography.net/forum/v ... php?t=1418
https://www.photomacrography.net/forum/ ... php?t=1472
http://www.photomacrography.net/forum/v ... p?p=249997

Some examples in:

http://www.photomacrography.net/forum/v ... hp?t=38761

http://www.photomacrography.net/forum/v ... hp?t=37800

http://www.photomacrography.net/forum/v ... p?p=241781

Best regards.

[apt403]

Regarding telecentricity - is there a guide you know of showing how to convert a normal telephoto lens to one with a 0 degree field of view? This might be a really great effect, lending a sort of scalelessness to the images, although if I'm not mistaken, it will mean I'm shooting everything at 1:1 essentially? It's hard to imagine what that even would look like.

You said I could add a 'limiting aperture'; I take it this mean introducing another optical element between the lens and the camera? And somehow this gives a telecentric projection?

I'm reading through all these threads now, this stuff is fascinating.

I do remember a thread documenting the journey towards a telecentric setup, but my search skills aren't bringing it up after a few minutes. I was just messing around with this approach the other day, so I took some quick snapshots of my experimenting.

The added aperture has to be at the focal plane of the lens - This is normally where the sensor would be. For Nikon's F mount, the flange focal distance (distance from the mounting flange to the sensor) is 46.5mm. I think the EF mount is something like 44mm?

I took your standard set of generic manual extension tubes, and stuck a paper aperture between the first extension ring, and the male bayonet. Then adjusted the length so that the total extension from the added aperture to the mounting flange on the lens was approx 46.5mm:






Note: I wouldn't recommend bits of electrical tape and white cardboard for a production setup, this was just some late night messing around on my part. There are much nicer, real irises that could be used.

I tested for telecentricity by taking a three shot stack, both with and without the external aperture, and had Zerene align the frames. Without the iris, Zerene reported a scale factor of about 1.02, or 2% difference. With the iris, Zerene reported 1.006 between frames. So, my setup isn't totally telecentric, but far better than before. If I had a few sets of extension rings, I might be able to get closer to the exact 46.5mm distance required. A bellows would also do the trick.

I haven't played with Helicon much other than messing around w/ the trial version, but in Zerene there's an option to turn off scale adjustments. This will eliminate the perspective changes that mess with stitching frames, but can cause ugly radial smearing in the individual stacks if there's too much change in scale between frames. With some higher higher magnification (eg. 10x) lenses, the change in scale can be so small that stacking w/o scaling the individual frames produces a nice result that also stitches without issues.

One can also use stacked lens to make a telecentric lens, as Rik has done here, for example (also a great informational thread, with links to more). I'm unfortunately not aware of the pros/cons between the two types of setup:

http://www.photomacrography.net/forum/v ... hp?t=39724

Regarding reproduction rate, I imagine w/ this technique your minimum would be defined by the ratio between the amount of extension needed to place the external aperture at the right distance, and the focal length of the lens. I pointed my setup at a ruler just now, and w/ the lens focused at the infinity mark (obviously not actually 'infinity' focus anymore) I got about 5cm in a frame. 5cm/23.5mm (APS-C sensor horizontal length) is about 1:2.13 reproduction. This is w/ a Tokina 100mm f/2.8 macro lens. Longer focal length lens will be less impacted by the added extension.

[rjlittlefield]

Fond as I am of telecentric optics, I'm not sure that they're the best solution for this particular problem.

The bad reputation of rotating around the entrance pupil, for macro applications, is because with classic optics like enlarging lens on bellows, the entrance pupil is someplace near the lens while the bellows and camera are far behind it. In that situation, it is mechanically awkward to support the lens+bellows+camera+rail in the right place to rotate around the entrance pupil.

But with modern closeup lenses, it's pretty common for the entrance pupil to be well back, sometimes very far back.

For example, my Canon 100 mm f/2.8 L IS USM lens at 1:1 has its entrance pupil very near the sensor plane. At lower magnification it moves forward, but it never gets beyond about the middle of the lens. True, the pupil moves around a little bit depending on focus, but for a subject like 617 wants to shoot, I expect that rotating around the entrance pupil would work fine with that lens.

I have no idea about the EF-S 60 mm, but if I had this problem, that's the first thing I would check.

--Rik

[617]

Fond as I am of telecentric optics, I'm not sure that they're the best solution for this particular problem.

--Rik

Rik, thanks for your input. I got a cheap pano head from Amazon and spent the morning dialing it in. I think I have it adjusted well enough.

See this test image:
https://www.easyzoom.com/imageaccess/51 ... 5bb5ed2a9b

Hardly macro by the standards of this forum, but a proof of concept nonetheless. In this set up, the lighting and subject are stationary and the camera is moving around the entrance pupil. This is a 2x2 tile with a good amount of overlap. I think this image is 6000 pixels in the short dimension, but after topaz sharpen / gigapixel AI, it can make this image even bigger while still maintaining good sharpness. I added a little film grain to this sample. I find that if gigapixel AI is used, the details have an odd painterly quality, and adding a little film noise helps them look a little more natural.

One interesting characteristic of this image is that it has a definite perspective, with taller objects converging towards the center. This is something I can manipulate easily by moving the art relative to the entrance pupil. This is something I wouldn't get with the telecentric set-up. If I wanted to flatten the image I think I could try using a 100mm lens, or just bring the camera back and take one picture instead of four up close.

I'm using two cheap video lights on goosnecks with no diffusers - they're 5"x7" panels about 10" above the art, so there is plenty of diffusion, and if I want sharper shadows I can move them out or use a flash.

This macro stuff is tricky to get right but the results are pure magic, even when taking images of visible-sized objects.

[rjlittlefield]

That looks pretty good. On quick scan I see one stitch error:


That could be due to parallax, or it could be something as simple as the stitcher not putting any control points in that area.

What software are you using to stitch?

--Rik

[617]

What software are you using to stitch?

--Rik

I'm using Microsoft ICE, totally automatic. I'll have to play around more to figure out why I'm getting those stitching errors. There's a good amount of overlap between the image. I'd prefer not to spend time making my panoramas stitch properly but it may be necessary.

I'm not sure if that's parallax related or not. I intentionally put those wires there because I knew they'd show any errors more readily than other objects. I suspect stitching software should be able to fix it.

[rjlittlefield]

Microsoft ICE, totally automatic

The good side of ICE is that it does a lot of panos quickly and well, fully automatically. The bad side is that when it does something you don't like, there's not much you can do to fix the problem.

If this result had come from PTgui, I'd say "Check the control points. Probably the automatic CP generator didn't happen to put any near that particular seam, so those tile edges are free to wiggle around under the influence of small errors far away. If that's the case, then manually put a couple of control points near the problem stitches and re-optimize."

But with ICE, no such treatment is possible. It doesn't even admit that there are control points, let alone provide any tools for manipulating them.

So I end up using ICE as sort of a quick-and-dirty first attempt. If the result is good enough, whatever that means in any particular case, then I'm happy because it was simple. Otherwise I haul out PTgui to see what went wrong and get it fixed.

Most of my panos are handheld landscapes with foreground extending inward to 20 or 30 feet. Your situation should be simpler than that, IF you can nail the entrance pupil location.

But given your workflow, there may also be other sorts of issues like slight changes in image scale due to different starting distances for the focus stepping in each tile. In PTgui, that requires selecting some non-default options that allow "focal length" to change between tiles. I have no idea what ICE will do in that case.

--Rik