[Feature] Provide Cover/Preview Image through local LAN

[chriswood1001]

Describe the feature

Can the cover/preview image be pulled into Home Assistant via. local LAN, not cloud? I'm referring to the colored rendering of the product that's shown on the A1 screen during the print, not the camera entity.

Bambu's credential push leads me to believe that read-only values should be available over LAN. However this particular cover/preview image feature appears to be reliant on the cloud.

What device is this for?

A1

Other Information

No response

[steveneaston]

I'm by no means a Python expert, but I've written a script that fetches the 3mf file from the printer's FTPS service using ftplib and extracts the primary thumbnail from it (if present).

To use it, download the bblab_fetch_image.py file from this gist and run it with the printer's IP, LAN-mode access code and 3mf model name (in quotes if it has spaces):

python3 bblab_fetch_image.py 192.168.1.123 12345678 "My super model.3mf"

It'll export the image file to ./images/My super model.png.

@greghesp @AdrianGarside I've got no idea how you'd go about implementing this into the plugin, or if you'd even want to but please use this script if it helps.

[AdrianGarside]

My concern is whether that would cause problems. Certainly early on after they introduced the FTPs support on the P1 printers it was always a gamble to download timelapses while printing as it might hang the printer (fans going, full temperature) in the middle of the print.

[AdrianGarside]

But doing it is potentially on the cards for if/when cloud access to the Bambu APIs becomes untenable to keep working. Or to reduce the take back if a bunch of folks are forced to switch to lan developer mode.

[t3chguy]

Might be good to have as an opt-in option anyway, I'd expect a 3mf file to be peanuts to transfer compared to a timelapse, and could be done during the initial heating/print readying.

[Woodcat64]

This would be welcome addition.

@steveneaston Excuse my ignorance. Could HA add-on AppDaemon or the pyscript integration run your script?

[steveneaston]

@Woodcat64 I tried using pyscript but I couldn't work out how to register the FTP_TLS class, my Python isn't so good. However, you can do it just using the shell_command integration instead.

---

1. Take my gist
2. Edit the following lines:
   (Replace models with www/bblab/models and images with www/bblab/images)

   -download_path = 'models'
   -image_path = 'images'
   +download_path = 'www/bblab/models'
   +image_path = 'www/bblab/images'

3. Put the modified script into your HA config directory
4. Add the shell command to configuration.yaml

   shell_command:
       bblab_image_getter: "python bblab_fetch_image.py {{ printer }} {{ access_code }} \"{{ model }}\""

5. In your template, use a Markdown card to render the image
   Edit: see follow up comment with a better solution for rendering the image.
   (Replace p1s with your printer)

   type: markdown
   content: >-
     ![Image](/local/bblab/images/{{ states('sensor.p1s_gcode_filename') |
     regex_replace(find='3mf', replace='png') | urlencode }})

6. Trigger the shell command using an automation, I'm using a change to the gcode_filename sensor.

   action: shell_command.bblab_image_getter
   data:
     printer: 192.168.1.123
     access_code: 12345678
     model: "{{ states('sensor.p1s_gcode_filename') }}"

[Woodcat64]

@steveneaston I got it mostly working. The issue is the markdown card does not show the image until I refresh firefox tab (didn't try the HA app yet). I will play with it when I have more time.

Thanks so much!

[coltography]

This along with filament length and weight would be great. I track those stats and so badly wish I could see them in LAN only mode.

[steveneaston]

The issue is the markdown card does not show the image until I refresh firefox tab (didn't try the HA app yet).

@Woodcat64 Unfortunately the markdown card isn't the best choice because it appears to not be reactive to changes. I also found it works in the companion app on the external URL but not the internal URL. With that, there's two options.

Simple option: Output a static file

1. Replace this line:

   -image.filename = f'{os.path.splitext(model_filename)[0]}{os.path.splitext(image.filename)[1]}'
   +image.filename = 'model.png'

2. Use a picture card

   type: picture
   image: /local/bblab/images/model.png

Better option: Use media-source

1. Set up a media source if you haven't already (documentation)
2. Replace www in the output paths of the fetch script (step 2 here) with the media-source name, e.g. media
3. Install the Media Source Image card from HACS
4. Add it to your template
   (Replace p1s with your printer, and /media/ with the media source dir you set up in 1.)

   type: custom:media-source-image-card
   entity_id: sensor.p1s_gcode_filename
   image: >-
     media-source://media_source/media/bblab/images/{{
     states('sensor.p1s_gcode_filename') | regex_replace('3mf', 'png') }}
   # Remove the following if not using a Sections layout
   visibility:
     - condition: state
       entity: sensor.p1s_gcode_filename
       state_not: unknown

[coltography]

Hi @steveneaston thank you for sharing all this. I'm trying to get this working and just troubleshooting by running the script directly in my ssh terminal and I get this:

python gcode_img.py my_printer_ip "my_access_code" Cube.3mf

Error: models/Cube.3mf is not a valid 3MF file.

although when I go into /www/bblab/models I do indeed have the downloaded Cube.3mf file. If I extract the file manually I can see the plate_1.png in the file so that parts working, but I just find it strange the script says its not a valid file?

I have also made the adjustments for /www/bblab/models and images folders as mentioned above.

Edit; Nevermind, I am blind. I didn't notice the (f'www/bblab/models/{model_name}' section also needed to be changed and only swapped the www/bblab/images part. It works now!

[coltography]

I added to @steveneaston 's gist and along with getting the cover_image.png it also can pull the filament cost, length, type and weight into HA input_number entities without using the cloud login!

https://gist.github.com/coltography/4b0d93e7d3376b8f875d5942a2f3d80b

Would be super cool to see this added to the bambu-ha integration somehow, since it just pulls the stats from the latest gcode file instead of pulling them from the cloud (or however it does it right now)

[steveneaston]

I've just made a change to my gist to fetch the image for the plate currently being printed. Orca/Studio bundles some XML as part of the 3mf with details on the current plate. I noticed when I printed a model yesterday from a second plate, it pulled the image from the first plate. The 3mf had both plates bundled, the correct image being plate_2.png

@coltography excellent work! You should be able to implement my changes easily into yours, since the gcode follows the same convention (i.e. Metadata/plate_2.gcode).

[drkpxl]

Seems possible to run this as an automation.. Something like

1. When new print begins as trigger
2. Run script
3. Save image url as variable to apply to HA card?

Anyone figure it out? I am deep in 2 other projects (Brick Layer and a HA Card) but seems reasonable to get to.