By following the guide up to the point of importing your excel results into Rhino, I came across is this software
My results ranged from 0.76 to 3.0, I subtracted my results from 4 (to get the inverse of my results)
Then dropped the decimal point and dropped the last number (if more than 4) or added a 0 (if less than 4, unless its less than 1000), to turn them into milliseconds "ms" xxxx
This way your only looking for the difference between the different numbers, and what the numbers are i.e. UV level or Time, it doesn't really matter
Dan
Following the guide my grayscale mask for an Optoma HD26
Mask Builder Settings
Same data values, but output by the Garage Science software
Seems to be pretty consistent, just need to smooth out the colour transfer from the different value points and have the ability to load and save settings.
Dan
Statistics: Posted by danthehuman — Sun Aug 13, 2017 3:10 am
]]>
I have to say, this is really great project!
Though, I have a small problem:
I've assembled my UV sensor and at the beginning I was getting some values between 13 and 17 but after couple of minutes of playing, I started to get values between 0.03 and 0.05. Did anybody had a similar issue? Is it possible that my sensor is already broken?
I have to mention that I was playing with normal bulb, my projector is coming in 2 day, then I will try to measure it again if nobody has any idea, maybe there is still some hope :)
Thanks for help!
Tom
Statistics: Posted by zet-jaro — Tue Mar 29, 2016 1:32 pm
]]>
]]>
]]>
]]>
]]>
I decided that I must figure out how to predict the resultant intensity for a given mask value and projector coordinate. So I chose 4 of the brightest points and sampled each 10 times at every gray value between 0 and 100, in increments of 5. The points I chose were 3, 5, 14, and 15.
I entered that data into a spreadsheet to get the averages: http://bit.ly/1M1AuWf
Then I produced this graph: https://output.jsbin.com/wojuro (X axis = black-to-white, 0-to-100)
If you're going to be measuring the UV output like this, you should get a pair of these (they work with glasses too):
http://www.amazon.com/gp/product/B0017Z ... ge_o01_s00
I'm trying to think of some good explanations for the noisy plateau at the top of each curve. Either the sensor is too noisy for this use, or the UV intensity is exceeding the acceptable range of the sensor (seems doubtful). Or maybe the noise is more apparent at full intensity, and the falloff for white happens to be very sharp. Thoughts anyone? My projector brightness is at 60%. What brightness are you using?
Not very comforting results, but I'm still hopeful I'll produce a map which is good enough. I still need to sample more points from the projection to see how dimmer areas respond to the mask.
Statistics: Posted by CaptainBisquick — Mon Mar 14, 2016 4:31 am
]]>
]]>
]]>
https://www.dropbox.com/s/5xhy1fei4rwia ... -0.2.blend
This is the file to produce a UV mask using Blender (open source, totes free dudes). The Blender interface isn't conventional, but if you do the UI tutorials, it's not that hard to use. But here's some of the basics:
• Tab key on a selected mesh and you can adjust the vertex data. The XYZ for the active point is in the sidebar of the 3D view.
• The 3D cursor is the thing that jumps to the mouse location when you left click in the 3D view. You can snap it to stuff (shift-S) and then use it as the reference point for transformations like scaling, if you set the 3D cursor as the pivot center (hotkey: the alphanumeric pad period key).
How to use this file:
• In the top right you should see a button that says "bake". If you click it, you should see an image render in the bottom right. In that panel, you can save the image to use as a UV mask (I recommend JPG).
• You use edit mode to manipulate the height map mesh. You then have to change back to object mode, so you can select the height map mesh, followed by the projection plane, and then press "bake", so you can save the image.
• Keep your height map Z values between 0 and 1. If for some reason your values are not in that range, you can use the normalize option. This just means the farthest Z distance gets mapped to either 0 or 1, and the projection plane is at 0.5 on Z. I strongly recommend not using normalize.
• I've raised one corner of the height map so you can see a bright corner in the output, and verify that your top and bottom are correct on the projector. I've also marked the bottom of the height map mesh with a green edge (only shows in edit mode).
Caveats:
(1) In CW (or at least the version I'm using), if you change the mask, you have to save it under a new name in order see the changes in CW. CW doesn't release the file memory for writing after it's been opened, unless you quit and reopen CW.
I don't have clear instructions for you on how to get the perfect output. I'm working on figuring out a method myself. The values of the image don't seem to have a linear relationship to the output UV value, which makes this harder.
The Z position of the projection plane is your 50% spot.
(3) I chose to extrude the edges of the height map mesh so that I would have a margin around the whole map. You can delete this extended border if you want, but I don't recommend it. The margin ensures that you get the expected mask values in the circles near the vat edges.
(4) This would be more useful if the resultant output values could be known prior to applying the mask. I'm working on this problem currently.
This method could be automated by writing a python script which imports all the CSV data and sets the height of the vertices for the user, and triggers the render process and opens the save image dialog. I don't have time to do this task, but if someone would like to take it on, feel free to increment the version and host it on dropbox. If the tool becomes polished enough we could put it on github.
Statistics: Posted by CaptainBisquick — Fri Mar 04, 2016 2:49 am
]]>
Slow PC will affect print quality.
Statistics: Posted by DanOrgan — Thu Mar 03, 2016 5:42 pm
]]>
It seems the problem was related to the use of PNG. I'm generating my mask image using displacement mapping in Blender (I'll share the blend file with you guys), and then any tweaks are done either in Blender before render, or in Gimp after render. Then I'm saving to a VirtualBox shared folder, since I'm running CW on VB. Somewhere in the process, the PNG format is getting jacked up. If I open the image that Blender is producing, I get the scan lines. If I export the image from Gimp, sections of my projection completely disappear. I switched to JPG and everything is working now.
Statistics: Posted by CaptainBisquick — Thu Mar 03, 2016 5:06 pm
]]>
]]>
Statistics: Posted by DanOrgan — Thu Mar 03, 2016 3:01 pm
]]>
Sometimes it looks like this (scan lines effect):
https://www.dropbox.com/s/x8qd5q5c0evrlwn/IMG_4389.JPG
And sometimes it's missing many rows of the image output.
Attached is the mask I generated. If you could try it and see if it works for you, that would be a useful test.
Statistics: Posted by CaptainBisquick — Thu Mar 03, 2016 7:09 am
]]>