focus steps, and other issues

[Tim M]

Greetings!

I’m trying to work up a new camera/lens configuration, and could use some help with what image “effects” are being observed, and what improvements might be possible, given the optics in use. I shoot Canon Powershots (fixed zoom lens) with a reversed 50mm f1.4 Canon FD prime lens. A Canon CHDK firmware hack script is used to step the focus of the Powershot lens, usually starting from around the minimum out, but stopping well short of infinity.

I image almost exclusively beetles (all from my backyard) and post over at BugGuide. You can read a little more about my setup and me here.

The current camera I’m trying to incorporate is a Powershot A650, with a 44mm (max tele) lens and a 1”/2.5 (7.6mm wide) 12MP sensor. I get about a 10mm wide field with the Powershot zoomed to 44mm. I’ve previously used an A640 (29mm tele), and a S5 (78mm tele), so I’m hoping this A650 nicely bridges the gap in field widths.

The image below is a 2.6mm portion of the underside of a Darkling beetle, Platydema ruficorne (Tenebrionidae). It’s a 8 image stack, using Helicon Focus. Among other issues, I think I might be using too course focus steps for this new lens/camera setup. Other shooting info includes f4.8, 1/125, and ISO 80. The full image and post can be seen on BugGuide here.



The CHDK script used looks like (partial):
@title A650v3
print "Shoot 598"
set_focus 598
rem sleep 2000
shoot
print "shoot 625"
set_focus 625
shoot
print "shoot 650"
set_focus 650
shoot
print "shoot 680"
set_focus 680
shoot
print "Shoot 710"
set_focus 710
shoot
print "Shoot 750"
set_focus 750
shoot
print "Shoot 790"
set_focus 790
shoot

. . . and so on.

To give some indication of focus step results, attached below are step #s 2, 3, and 4.





I probably should just go back and redo the script with many more (finer) steps. But as much as BugGuide has helped me learn beetle IDs, I thought this wonderful forum might help take my image efforts beyond my standard trial and (mostly) error approach.

Thanks, and comments / suggestions welcome
Tim

[rjlittlefield]

Tim, I agree with your suggestion that you "should just go back and redo the script with many more (finer) steps". When I look at the images you posted, they appear undersampled by a factor of at least 4X, quite possibly more. For calibration purposes, I suggest to have your script step 1 instead of 25, then see in stacking how many you can skip while still getting the resolution that you need.

It's worth noting that the way you are stacking lenses, the required "step size" doesn't depend on actual magnification, only on camera settings. What happens is that the added lens essentially enlarges and relocates the subject to a virtual distance farther away, and then the camera steps focus through that virtual subject as if it were a real one. If you stick different lenses in front, you will get different magnifications, but the camera will deal with the corresponding virtual subjects in the same way.

--Rik

[Tim M]

Ok, I was able to knock out a script with steps of 5 vs. 25, at least through the majority of the script. The results below appear at least improved in many areas where previously there was sigificant blurring. (the thin white lines are from some stray glue when I pointed the beetle)



Even with step sizes of 5, I can still see changes in each image as the stack is processed. Rik, I didn't ignor your request to go to step sizes of 1, but I realized it will take a bit more work just to try it. You can see my scripts are "brute force", not in do-loop form which would make changing the step size much easier. I used the former programming technique because typically I had scaled up the step sizes as focus moved from min to max, and when using < 30 steps it was easier just to specify each step. Also, I can get to steps of 2 on my 2GB cards, but to go to steps of 1 (>300 images) I'll have to figure out how to get CHDK onto a 4GB card (the formating procedure is more complicated and I haven't tried it previously). May take another day or two.

All this also makes me think I need to go back and re-do the scripts for my S5 (which has a larger tele lens). I thought I had them working, but now not so sure.

Comments welcome

Tim

[rjlittlefield]

Tim, this new image looks a lot better.

There are a bunch of different programming styles. I'm not familiar with CHDK, so when I saw your straight-line script, my first thought was "program generator". Then after a quick thought about the tools I'm good at using, I said "Excel". If this were my problem, one efficient way to handle it would consist of an Excel workbook using spreadsheet functions to construct CHDK commands. Then I'd just copy/paste from the spreadsheet into whatever text file it takes to drive CHDK. On the other hand, for someone not handy with Excel, that would be a lousy approach. I'll have to leave it to you to decide what's a good approach in your situation.

Regarding size of the memory card, it seems unlikely to me that your optics really justify full resolution. 2 GB divided by 300 images would be 6+ megabytes per image, which seems like plenty assuming that you're saving images as JPEG. To run the test, I suggest setting the camera resolution to whatever lower setting allows all the frames to fit onto the card. Then after you figure out what step size is appropriate, you can go back to the larger image size if it seems to add quality.

--Rik