RaspberryPi HQ and Fujian 35mm f1.6 deep dive
In my last blog I talked about the reasons for building an OpenScan classic with an RPi HQ camera. I also showed a bit of the increase in quality and sharpness of the scans.
I sort of glossed over the whole project because I knew there was no way of including everything I want to talk about in one blog post while also keeping most of you reading until the end. That’s why, very early in writing the first blog post, I decided to make a series of articles. This is the second article, a deep dive in the pro’s and con’s of the 35 mm 1.6f Fujian lens!
What constitutes a good photogrammetry image set?
Now, let's take a few steps back. What constitutes a good photogrammetry image set for say a miniature? Most articles many of us have read state something along these lines:
A dark or contrasting background
Enough depth of field to have the model in focus, but not too much to keep the background out of focus.
Use the highest resolution possible.
Each point of the model surface should be clearly visible in at least two high quality images. On OpenScan 150 to 200 images usually suffices! Any more just means more processing time. Anything less could mean loss of detail, but your mileage might vary.
Always move around the object circularly when taking photos.
Do not change view point more than 30 degrees.
Many of the items you’ll find online are covered by the OpenScan hardware, and software, and are not part of these blogs. The ones we care about right now are 2, 3 and 4.
Highest resolution, as I talked about in the previous post, is one of the main reasons for doing what I’m doing. Getting most of the object I want to scan onto the camera's sensor!
Item 2 and 3 however, getting two or more high quality pictures of every part of the model and getting a perfect DOF, are not as easy as they seem. There are so many variables that come into play, when using the setup I’m using, that affect image and thus mesh quality, they are worthy of their own time in the spotlight. I’m not talking about setup of the scanner (though that is very important) but things that affect image quality, for example:
Distance between object and camera
ISO (though we have no control over that within OpenScan software)
Use of Macro extension tubes, filters, macro lenses, …
Sensor size/crop factor (though honestly, I know little about this)
What is DOF?
DOF, or Depth of Field, is a range of distances from your lens outward that are in focus. With a large DOF the entire picture is in focus, this is often used in landscape photography. Narrow DOF is often used in macro and portrait photography to isolate the object from the background. When using a specific lens/camera combo, you can only manipulate DOF by changing the aperture. Lowering the aperture number opens the iris in the lens. Increasing aperture numbers closes the aperture in the lens.
A picture speaks a thousand words, so here is a visual representation of DOF.
Image credit: https://photographylife.com/what-is-depth-of-field
DOF encompasses a lot more than what i'm covering, if you want to know more, this site explains nicely!
Another thing that affects DOF is using macro extension tubes. Having one, or more, macro tubes means your MOD, or minimal object distance, lowers significantly. This allows you to physically move closer to an object and still be able to focus. It also means your maximum focal distance decreases from infinity to… well not infinity and your DOF will also get more and more narrow.
The Fujian 35 mm f1.6...
When looking for proper lenses to use with the OpenScan classic, I was overwhelmed with a lot of technical details. I know the basics from when I learned how to use my DSLR, but nowhere near enough to know what kind of lenses would work with photogrammetry and how macro extensions and different image sensor sizes would affect image quality. Even after reading many sites and explanations, I still found it hard to extrapolate that to photogrammetry setups, where you want to be up close and personal with the model. I chose the 35 mm Fujian pretty randomly based on some reviews online. These are the "Features" stated on the seller's website:
Mount type: C mount Focal Length: 35mm Aperture: F1.6-C Angle of view: 18 Aperture blades: Four groups Four Blades Frame:APS-C lens Coating:MC Multilevel Lens Size:37mm Iris & Focus Operation: Manual Minimum Object Distance (M.O.D.): 30cm Focus: focus to infinity Image plane size (Format): 1 / 2 inch
Now, these stats should be taken with some huge grains of salt. Especially the MOD is not 30 cm but 50 cm, and the sensor format is NOT 1/2. If used on my Canon, I will definitely get vignetting at some point. However, it's still a great cheap lens.
Testing MOD (Minimal object distance) and DOF!
With the extension tubes ready, I started experimenting. At first, I tried using the lens without any extension tubes. Well, technically I used one, but that was needed because it's a C lens on a CS mount camera.
I could not focus on the model at 24 cm distance, but the RPI HQ does have a "backplane focus ring" that acts as a variable extension tube. The BFL, or backplane focal length, is intended to overcome differences in lens designs. It can be extended to about 5mm before you don't have enough threads to keep is securely in place. I wanted to know if I could get the lens to focus at 24cm without a macro tube. The lens was able to focus when the backplane focus ring was extended about 3mm. Below is a table of all the tests I did with a superglue bottle cap.
The only variables for each set are the relation between aperture and exposure.