bvf Posted April 16, 2021 Share Posted April 16, 2021 Good ideas are there to be stolen ... I was absolutely intrigued with Any Perrin's post on Flow Visualisation ( https://www.ultravio...__fromsearch__1 ) and just had to try it out. But as I didn't have access to the software used to create the visualisation from the two images I needed a simpler approach, and tried simply taking the diference betwen the two images. Long story short, this works to some degree, although the image quality is best described as grainy (like the old Kodak 2475 Recording Film with push processing, for the old timers here). Some examples below. Post processing was mainly based on the histogram, bringing the sliders in (esp. the high-end slider) and moving the mid point to make the air flow visible. I tried making some stereo shots, but could not get any good results. Image quality is probably too low, and anyway a stereo image probably doesn't make sense here as you are looking through the subject rather than at its surface. My only tool for creating the screen was Excel + VBA. This could provide a screen with 200,000 random dots: this took hours to run, but worked in the end. I printed 2 copies on A3+ photo paper, and for some shots taped these together to create a larger screen. I also tried a far quicker method using Excel's in-built dorreddotted-line feature. This worked, but caused problems when viewing the results because of moire pattern interference with the pixel array of the monitor, so I stuck with the random dot screen. In keeping with the principle of stealing Andy's ideas, first some candle images. The next image was created using the dotted line screen: The next image also used the screen, but some blurring was added to try to reduce the moire effect: And just so I can claim to be on topic for this forum, here is a candle using UV (Baader U). I also tried IR, but couldn't get an image - I think the problem was that the ink used to print the screen was largely transparent to IR. Moving from hot to cold - here is ice on a flat metal spoon with holes in it (I'm sure it's got a proper culinary name): Back to heat, and a couple of shots capturing the output from a hair dryer. The heater was switched on - I could not capture anything with just a cold air flow. At last, time for a cup of tea: Here is the kettle spout. There was visible steam coming out of the kettle, but it didn't have any of this detail in it: Link to comment
Andy Perrin Posted April 16, 2021 Share Posted April 16, 2021 Haha, Bernard, subtraction is actually recommended in the research literature as a method of previewing BOS shots. You can download software for free like http://www.openpiv.net/ or https://eguvep.github.io/jpiv/index.html if you want the real thing. The former is Python and the latter is Java. Both have versions with a GUI. "What one monkey can do, so can another." - Richard Feynman If you decide to stick with subtraction, a better approach is to print a black->white or vice versa gradient instead. When you do the subtraction you will get a less grainy image. This has even been done with an LCD screen as the background... Link to comment
Andrea B. Posted April 16, 2021 Share Posted April 16, 2021 "This gets interestinger and interestinger." - Andrea B. Link to comment
SteveE Posted April 16, 2021 Share Posted April 16, 2021 Could you make the random dots by putting black paint on a toothbrush and scraping your finger across the bristles causing a spray of fine droplets to be projected? When I was a kid we made "shadow" hand prints of our hands using this technique with white paint on black paper for Mothers Day gifts. Link to comment
Andy Perrin Posted April 16, 2021 Share Posted April 16, 2021 SteveE, you could, but you would have no control over the dot size or density, so that's probably not a good method! It's much better to print them. Either write a script to do it or use one of the many BOS pattern generators already out there. Link to comment
colinbm Posted April 17, 2021 Share Posted April 17, 2021 Fantastic Bernard, some show a real 3D like view, especially the last cup-o-tea. Link to comment
colinbm Posted April 17, 2021 Share Posted April 17, 2021 The potential is infinite.... Link to comment
Andy Perrin Posted April 17, 2021 Share Posted April 17, 2021 Colin, that’s the mirror-based Schlieren. The detail you get by that method is incredible but you are limited by the size of the mirror and the size of the room. To get even a 130mm telescope mirror with a focal length over a meter (which you need) seems to be around $100-150 range. I might still try it anyway... There are a whole family of Schlieren methods, single mirror, double mirror, shadowgraph, BOS... Of these, BOS is the easiest and cheapest, followed by single-mirror like the Veritasium YouTube. Link to comment
bvf Posted April 17, 2021 Author Share Posted April 17, 2021 The potential is infinite.... There's some very neat stuff there, with great quality and the ability to capture fine effects. As Andy identifies, its limiting factor is the mirror size. The screen + differences method is limited, but it's so easy and doesn't need any special kit. Within a day of reading Andy's post I had my first image. And it really is gobsmacking when you overlay your first pair of images and select difference mode and the flow image emerges. Link to comment
bvf Posted April 17, 2021 Author Share Posted April 17, 2021 If you decide to stick with subtraction, a better approach is to print a black->white or vice versa gradient instead. Can you explain what you mean by this, Andy? Just a simple gradient from black to white from top to bottom (or left to right? or top-left to bottom-right?) - nothing more complicated than that? Do you actually get any image in the near-black and near-white regions? Link to comment
Andy Perrin Posted April 17, 2021 Share Posted April 17, 2021 Bernard, I sent you the paper I read on this but I haven’t tried it. They do say very close to the edge you get bad data. So don’t do that! The paper is on using color in fact — you make a gradient in each color in each direction so you can tell vertical density gradients from horizontal. Link to comment
Recommended Posts
Please sign in to comment
You will be able to leave a comment after signing in
Sign In Now