A journal following the history, design, construction and operation of Bernard Kempinski's O Scale model railroad depicting the U. S. Military Railroad (USMRR) Aquia-Falmouth line in 1863, and other model railroad projects.
©Bernard Kempinski All text and images, except as noted, on this blog are copyrighted by the author and may not be used without permission.
-----------------------------------------------------------------------------------------------------------------------------------------------------------------

October 23, 2018

Initial Impressions Arsenal DSLR Intelligent Assistant

Final image using Arsenal captured images and Photoshop image Stacking
Test set-up with Canon 70D and  Arsenal mounted on hot shoe
Setting up the Arsenal is tricky.
I got this error message quite frequently

Background


Model railroad photography is a unique branch of photography.  If you take a conventional professional photographer and ask them to shoot a layout, they probably won't get great results. They will get well focused and exposed shots, but they probably won't look real. It takes some experience to learn the tricks on how to make model trains images look real.

Why? We are trying to take macro range images of our models and make them look like natural fmages taken outdoors where everything is in the hyperfocal region. I discussed this before here.

The use of focus stacking software has changed how we achieve that objective. Instead of using one image where we hope everything is in focus due to a tiny aperture, we use multiple images where only a small sliver of the image is in focus. We then use image processing on a computer to create one sharp image from the set of in-focus slivers.

The process of taking the slivers can be tricky as you must manually move the camera focus in a step-wise fashion to get complete coverage of the scene. If you go too far on any individual step, you may leave a portion of the image that is out of focus. This is made more difficult if you use a wide aperture for each sliver, as each sliver with have a smaller section in focus.

In using focus stacking, I find I get the best results with apertures around f9-f16 and 6-10 images. It depends on the scene. Photos that will have a lot of immediate foreground need more images, as the foreground slivers are narrower than the slivers in the background.

The Arsenal 

Arsenal is processing the stack of images.
This can take 5-20 minutes
Once connected this is what the live
view on the Arsenal looks like
When I learned about the Arsenal intelligent assistant about two years ago, it sounded interesting. The automated focus stacking especially caught my attention. So I was an early Kickstarter adopter. The device arrived this week, and I did a series of test shots.

I followed the directions and let the device charge over night. Meanwhile I downloaded the Arsenal Ap to my phone.This Ap will control the Arsenal device, which I mounted on the Canon 70D camera.  Once everything was mounted up  I installed the latest firmware updates and got the Ap to connect to the camera.  The Ap creates its own WiFi network. The connecting  process was finicky. It took me several tries to connect the device to the phone. It appears that the USB plug is very susceptible to improper connection. Once I got it connected, I only lost connection once in a two hour photo shoot.

I shot a series of images using the focus stacking feature. The Arsenal does a pretty good job of advancing the focus to get the slivers. However, it seemed like the Arsenal had real trouble in taking up close slivers using my Macro lens. So I switched to my 18-135 zoom and it worked better.

As the Arsenal makes the camera take an image, that image is written to the  SD card on board the camera. Once all the images of the stack are taken, it loads them to the Arsenal Ap via WiFi. But it does not load the full res images to the phone. So once you end the session, the high res images are erased from the Ap. However, they are still on the SD card in the camera.

Then the Arsenal tries to create a single image from the stack of images. This process takes anywhere from 5-20 minutes. In some of my test cases, it never finished processing and I did not get a finished image. When I did get a finished image, I saved it using the share button on the Ap. I saved mine to the phone, but you can email or send to social media.  The Ap stops processing if the connection to the camera is broken. That is annoying. I'm not sure if it picks up again when you reconnect. I don't think it does.

In general, the processed images from the Arsenal where not usable. Below you can see three of the results from about 13 test images. The stacking processor in the Arsenal is not as good a Photoshop or Helicon Focus.  It's slow and doesn't always give a fully sharp image. Sometimes the processor hangs up and never gives a final image. That's the bad news.

The good news is that the stack of images are on the SD card. So I imported them to my computer and used Photoshop to get the final stacked image (see lead image above). That image looks fine.

The Verdict

My initial reaction to this device is mixed.

The wireless connection was finicky to set up and will cut out if you make any camera changes or jiggle the USB cord.

The device does a good job of taking the images for the focus stack, but the on-board processor was not able to consistently create a final image.  If you have a recent version of Photoshop or Helicon Focus you can use the stack that the Arsenal generates to get a good image, but you won't be able to rely on the on-board processor to handle it for you. I'll do some more tests and report the results in the future.

The Arsenal is capable of other features like HDR and time lapse, but I did not test them.  I did not take the device outdoors, so I can't comment on how it works in that environment.

Test 1 looks like the focus stacking processing was not complete


Test 2 was one of the better of the Arsenal processed images, but still not as good as Photoshop focus stacking

Test 3 was a success. This image doesn't have as much detail very close to the camera. The Arsenal processor was
able to make a nice sharp image from this test. I tried this with 10 images and the processor got hung up.
But with 6 images the Arsenal was able to process the stack.


No comments:

Post a Comment