trying to get data out of marimapper into xlights or wled

[swinokur]

Hi there,

I've been successful in using marimapper with the wled backend to get a scan of a dome of 2915 LEDS:

And I have the CSV data - but I'm a little stumped as to the next steps - I'd like to get the model into WLED (ideally) or xLights (2nd best).

wledmap.zip

attached are the csv files.

I did try the python script that's here: #23 (comment) but the points come in at a weird angle and that tool doesn't appear to give me a way to "relevel" the dome.

I also found this script: https://gist.github.com/joeycastillo/1f6de9d291d85fe1f1714a6aeefc1fe4 which let me convert to a pixelblaze map which I could then preview here: https://jasoncoon.github.io/led-mapper/

But those don't seem to help me get data back into wled or xlights (but at least its a good check that "led-mapper" did show me a dome!)

I was wondering if you might have some tips/tricks/tools to try? btw, the scan was performed with a logitech 920 camera, and the default camera in the software. I did try the opencv_full camera model first and wasn't able to get a successful scan with that one - the software just seemed to decide that it wasn't a full hemisphere!

thanks!

[TheMariday]

Hi there!

That's an incredible looking dome you've got! I really hope that I can help you get this into xlights.

From what I understand the issue here is about orientation, and for that I do have a hack!

Marimapper uses the first camera position as the primary coordinate frame. So one thing you can do is this:

• add a new scan with the camera perfectly perpendicular to your Dome.
• close marimapper
• swap the CSV file names of scan 0 and your new perpendicular scan
• re-launch marimapper
• profit!

This is actually one of the main reasons I am refactoring married mapper at the moment to allow you to reorient models, however that's still a few months off!

Please do let me know how you get on and thank you for raising this issue!

TM

[swinokur]

Hey there,
Thanks for the quick reply! I will give your suggestion a try. I guess I didn't realize that the tool would "pick up" previous CSV files when run - I thought it was a single shot tool? Is there a command line needed to get it to regenerate the 3d map from the 2d scans?

I guess the other part of the question was - and I guess this is the tricky 3d to 2d bit - WLED only supports 2d meshes so I guess I am going to need to learn how / figure out / find a tool that can turn the 3d csv data into something that will work in wled as a 2d mesh... and I guess create the 2d json file that wled expects? (I imagine that's more than a single step)

thanks!

[TheMariday]

You're more than welcome!

All you need to do is launch marimapper in the same directory as the 2D CSV's and it'll auto-regenerated the 3D.

Flattening to 2D will require some code.

Here is a //very// quickly slammed together snippet if it's of use that should quantise your 3D map into a 100x100 grid along the z axis.

The main thing to consider here is the fact that as we're reducing it down to a grid, multiple pixels may try and map to the same Square, in this case only the first mapped pixel will be able to be mapped.

This is one of the reasons why I haven't offered any quantization tooling because it's quite a confusing problem to have unless you can do something really smart when it comes to slightly moving pixels around.

Let me know how you get on and also looked at your scans and they looked fantastic!

import json
grid_x = 100
grid_y = 100

grid = [-1 for _ in range(grid_x * grid_y)]

with open("led_map_3d.csv") as file:

    lines = [l.strip() for l in file.readlines()[1:]]

    indexes = []
    x_vals = []
    y_vals = []

    for line in lines:
        index, x, y, z, _, _,_,_ = line.split(',')
        x_vals.append(float(x))
        y_vals.append(float(y))
        indexes.append(int(index))

    x_vals_norm = [(x - min(x_vals)) / (max(x_vals) - min(x_vals)) for x in x_vals]
    y_vals_norm = [(y - min(y_vals)) / (max(y_vals) - min(y_vals)) for y in y_vals]

    x_vals_int = [int(x * grid_x) for x in x_vals_norm]
    y_vals_int = [int(y * grid_y) for y in y_vals_norm]

    for i in range(len(indexes)):
        try:
            grid_index = y_vals_int[i]*grid_y + x_vals_int[i]
            if grid[grid_index] == -1:
                grid[grid_index] = indexes[i]
            else:
                print(f"failed to map LED {indexes[i]} as map square already populated with LED {grid[grid_index]}")
        except:
            print(f"failed to map index {indexes[i]}")


    json_map = {"map": grid, "width:": grid_x, "height": grid_y}

    json.dump(json_map, open("wledmap/ledmap.json", "w"))

    print(json_map)

[TheMariday]

Oh side note, your perpendicular scan doesn't need to be a full one, just enough for the camera to reconstruct!

[swinokur]

The rescan came out fine!

I think the challenge is that this is wired up as 4 (5 if you count the logo) 1d strands of LED - so a matrix mapping won't really work. I actually need to have something convert from 3d to 1D, really.

That might be a job for ... touchdesigner? (have you ever pulled your csvs into Touch?)

I did point LEDfx at it earlier today, so its nice to see all the LEDs lighting up when I send a bunch of DDP packets at it.

[TheMariday]

Sadly I haven't worked much with touch designer. So effectively you're looking to sort your LEDs into one list based on some criteria?

It should be quite simple to do depending on whether you want them ordered top to bottom or around the edge, etc etc.

[swinokur]

So looks like TouchDesigner can import the CSV directly, now I just need to figure out what to do with it in touchdesigner!

[swinokur]

okay, I've gotten a bit further and have it sort-of working in touch designer.

The main issue seems to be that the data points don't seem to be sitting "flat on the table". Here's how touch interprets the 3d csv:

I tried manually entering some values into touch designer's transform, but it didn't get me that close.

I'm wondering if you might know of a tool that will let me transform all of the mapper's points so that the dome is sitting flat on the z axis?

[TheMariday]

Sadly not off the top of my head.

Did you swap the scan_0 csv as outlined above for the perpendicular scan? That should have straighten it out.

Either way I wish I could time travel and get you Marimapper V4 as it would be trivial with the new version.

Sorry I can't help further on the orientation side of things. It's come up multiple times and hence why I want to make transforms easier with v4.