Better diffusion

[SuperDave42]

Awesome photos, that really helps! The density-concentration approach I took is to mimick designs for LGPs I've found when researching how they are usually made, but I hadn't seen the Nanoleaf one up close until you posted it. Perhaps we don't need to compensate for the drop off so much if we're lighting it from 3 different angles at once.

Jim

Seeing your density concentration pattern really intrigued me and is what motivated me to go out and buy the Nano to tear down for a better look.  Variable density is not needed when evenly lit from 3 sides facing into the center.

How did you create your pattern ?  I'm interested in that for other fun stuff.

[NotionalLabs]

My pleasure, your super detailed teardown was fascinating - I appreciate you spending the time and money.

To create the pattern I wrote a small script in Python that uses the PIL (Pillow fork) library to generate the high-resolution 8-bit greyscale bitmap for raster engraving, and an output mask that I used to both mask off the final output dot pattern (like I posted further up) and save as a standalone image that I could use to generate the vector cutting path in Inkscape (via the "trace bitmap" feature). The pattern itself was generated by playing with a parabolic quadratic equation until I was happy with the shape of the curve, then using the output of that to give me the appropriate dot spacing for a given X and Y line. This gives a very regular spacing though, so I then calculate a random variable to offset the X & Y coordinates of the actual dot placement each time, which generates the more diffuse pattern that I posted. I thought this might be serviceable for a first try (especially since I haven't engraved an LGP before so I wasn't totally sure if it would produce much of an effect without professional tools or a lensing surface, etc), but one big issue is that the pattern still has a bias towards the cardinal "north/south" and "east/west" axis. That's not ideal for a triangular LGP with 3 opposing entry points, so I was about to rewrite the code using some trig to calculate the dot placements (more like a circular gradient approach) when I saw your post

I'll give a regular pattern a go, your photos and description are perfect - if I get a workable script together I'll post it so you can hack it to whatever you have in mind.

[NotionalLabs]

I managed to get on the laser today and I'm super impressed with the results! It's very hard to photograph in a way that's representative of what it looks like in person, but the diffusion with a regularly spaced LGP is significantly better:



Vs a stock Moodlite v2 without any LGP or changes (printed in white PLA, 20% infill):



The LGP is based off SuperDave42's teardown, with 0.5mm dots spaced 1mm apart, engraved at 55% power out of 3mm clear acrylic. I'll clean up my python code and the images used to create this and post them up here for folks to try if they have a laser engraver handy. Here's a photo of it in-place:



The photographs don't really do it justice, the difference is huge and the bright spots in the photos aren't anywhere near as pronounced in person. Here's a video of it in action: https://imgur.com/HFMMoxq

Jim

[SuperDave42]

AWESOME! looks a LOT better

I just picked up a sheet of this material during lunch today
It's similar to the top sheet in the NanoLeaf, just a shiner smooth top surface.

I also notice that other LGP makers use the Dot Matrix pattern as well as a Line Matrix pattern   (grids)    I suppose honeycomb matrix patterns etc could also be used.

[NotionalLabs]

Hey all,

Here's the code and source images that I used to engrave the panels in my photos: https://drive.google.com/open?id=1EvXxb8voaSalf7lAFy04ohaIDxxQmt81



The pre-made pattern, mask, vector path, and the exact Adobe Illustrator file I used to create it are included, along with the Python3 code used to generate the images. I quickly annotated it in case it's helpful for anyone who wants to tweak the parameters of the LGP for their builds.

I briefly tested using some folded aluminium foil to black out the corners and it seems to improve the diffusion appearance a little by limiting the bleed through the 3d printed top:



@SuperDave42 - I'm looking forward to seeing your tests with the proper diffuser material, that could take the build to the next level! The edges of my LGP are wide enough to fit 3 LEDs to a side using the higher-density light strips we discussed in the other thread - I think I'll make some small edits to the moodlite v2 model to accommodate that arrangement and post the results.

[Allistar]

https://www.reddit.com/r/prusa3d/comments/aj9p3a/i_designed_my_own_scattered_rectilinear_infill/

Look at that amazing diffusing happening with this guy's developed infill option...  Has anyone tried this yet?

[hauschka]

Definitely some great work.

Would be interesting to see how it looks like on the moodlite panel.
However, his box is massively big which definetly helps with scattering the light as well!
(as in...there is a long distance between LEDs and the diffuser)

Regards,

[thenickdude]

The scattered-rectilinear infill just makes the infill pattern less visible on the surface, I don't think it contributes much to the actual diffusion (diffusion in my design was provided by 1x0.1mm white bottom and 1x0.1mm white top, the total diffuser thickness is 5mm).

[DevilsDJ]

https://www.reddit.com/r/prusa3d/comments/aj9p3a/i_designed_my_own_scattered_rectilinear_infill/

Look at that amazing diffusing happening with this guy's developed infill option...  Has anyone tried this yet?

That guys leds are on the bottom not the sides and the distance between bottom and diffuser is bigger than with the moodlite. For me I made a 5mm taller top with pla diffuser (just remodeled the original top to be 5mm taller , moved the pins and made the sides a bit thicker for less diffusion on the sides) I am thinking of remodeling the tile top frame only so I can fit a 2mm acrylic which I will be sandpaper frosting with the top tile raised 5mm again . Will post results when I do.

Note: this is my raised tile top , I'd say it is much better for diffusion then the original , you cant see the pyramid almost at all. The photos dont do it justice , looks better in real life

[tg44]

Is sb tried out non-clear plexi? Maybe non-clear plexi with some bathroom glass coverer? I just surfed on the internet and find some really interesting plexi, just wanted to ask if anybody has any experience with them. Also what about layering multiple diffusion layers?

BTW what do you think about the square and hexa shapes (fix the diffusion with picking other shapes)? I think the hexa is much easier to light up without "blindspots".
Other question: Is sb tried to place leds to the edges?