Ambrosinus toolkit

[beardface00 (David)]

Hi Luciano @Ambrosinus , thank you for your efforts in developing these tools! They are amazing!

I am hoping to install this plugin on a laptop within my organisation, but think I will run into some permission issues.

These are the settings I currently have which work for SD (locally) and Rhino. Can you tell me if these settings will work as of today? Or whether I need to ask my administrator for access or updates?

• Rhino v7 SR34 (7.34.23267.11001, 2023-09-24)
• Stable Diffusion v1.5
• ControlNet v1.1.440 (models from lllyasviel/ControlNet-v1-1 at main)
• Ambrosinus Toolkit 1.2.6 (Package Manager)
  - Cannot run Rhino as an administrator!
  - Updates to the Automatic1111 installer are run centrally by the administrator as well, so that may result in the “webui-user.bat” file being overwritten or bypassed upon every computer restart.

Also, should the “value list” components automatically “populate” with the correct values when plugged into CNVers, Mode, Resize & Sampler? (as seen in your images on Ambrosinus-Toolkit/AI_components/SD_ControlNET_locally at main · lucianoambrosini/Ambrosinus-Toolkit · GitHub)
Or should they be manually entered? I haven’t been able to tell from your videos.

I have not been able to open the demo files (i’m guessing because of the rhino version or administrator access?), but get Grasshopper IO errors when trying to open saying “IO Read error: Unknown data type encountered.”

Apologies for asking so many questions! I’m just excited to get started with your tools!

Thank you again!

[murgodman]

@Ambrosinus there is an installation error after following your installation guide from Running Stable Diffusion and ControlNET locally via Grasshopper (thanks AUTOMATIC1111) - YouTube

I faced following error

please help to resolve this

[Ambrosinus (Luciano Ambrosini)]

Hello and thanks for your message.
What do you mean about installation? It is more likely that the issues concern something about Python libraries installation. I see that you are using a python virtual environment. I tested the toolkit for a standard use of python, and I do not recommend using Anaconda (but maybe it is not your case).
please could you describe your procedure in detail? Thanks.

[murgodman]

Hi Luciano,

Thanks for the reply. I installed python 3.10.6 like mentioned in C drive and I get hugging face SSL error even though I installed their certificate in Trusted root directory and added cacert.pem with their respective certificate alphanumerals. The issue still persists.

I also could load stable diffusion checkpoint here. Even if I choose from drop down list it throws error in command prompt saying SSL error from hugging face. Please help I have already tried for 2 days

[Ambrosinus (Luciano Ambrosini)]

Hello and thanks, but I want to understand if the problem is triggered when you use the Grasshopper toolkit. and if yes on which components do you get this error? …Until now no one talked about any SSL certificate…so it is weird, and seems a problem related to Python and the machine.

I should rebuild firstly your approach, but basically no SSL certificate or Huggingface was mentioned in my video…so I say so weird until today . it seems related to Automaticc1111 installation. We can move the conversation by email if you want

[murgodman]

Hi Luciano,

sure please share your mail id

[Ambrosinus (Luciano Ambrosini)]

Sure, please check it out there (Food4Rhino page)

[Vasudha_Karnani]

Hi Luciano! @Ambrosinus
Thankyou for this plugin!! I was hoping to get some help on identifying my issues of using ControlNet via this plugin.
some info - i have python 3.10.6
Installed SD v1.5 from atomatic111 like your video.
I tried using Controlv1.1…didn’t work so removed the entire thing, reinstalled Stable diffusion and Downloaded ControlNet1- archived as per your resources…

my issue is coming while launching the bat file. im getting API errors as well as some others while it tries to load CN. It work if i switch to just T2I basic mode…, but if i try to use Canny or or some other CN model, i get errors like - out of range, or error - ‘1. Solution exception:Index was out of range. Must be non-negative and less than the size of the collection.
Parameter name: index’

PLease help me identify how i can make Control Net work Properly! Also attaching screenshot of errors in bat file! Thanks a lot!

[Ambrosinus (Luciano Ambrosini)]

Hello Vasudha,
thanks for your interest and tour inquiry.

1. Can you run stable diffusion (A1111) web-based app? I mean without the toolkit.
   Because this toolkit has some components that create a sort of gh-interface for SD by A1111, we need to be sure that you can run it before the toolkit. If not, something was wrong during the installation of the A1111 client.
2. I suggest installing in a folder the A1111 project and then the CN (latest version) from here: Mikubill github.com

here is an old video Demo about how to install A1111+CN (with CN v1.0 and CN v1.X, I recommend the latter one, take the link from Mikubill link) How to install AUTOMATIC1111 with ControlNET v1.0/v1.X - YouTube

I have seen a different address; I mean this client works as the default settings in a localhost 127.0.0.1:7860 but I saw 0.0.0.0:7868.

First of all, please, be sure that you can run properly A1111 + CN project properly by the web-based App, after this we can investigate the issues related to the toolkit.
Please, let me know.

[VAT]

Hi Ambriosinus,

I’m facing the issue as shown in the image. I have followed the steps as mentioned in your blog. Please let me know how to resolve the error.

[Ambrosinus (Luciano Ambrosini)]

Hello Vinod,
thank you for your inquiry and interest in the toolkit!

For further assistance, I recommend sending a private message with additional details such as your Python version, toolkit version, and whether you’ve used the GH demo file from my GitHub page for initial tests. If possible, you could also share your GH test file with me. ASAP I will attempt to provide you with an answer or proceed with troubleshooting steps.

Thanks!

[Chris_C (Chris Caceres)]

I gave up trying to run it in Rhino / Grasshopper. You can do it the other way by running SD + Control Nets in ComfyUI and screen-grabbing your Rhino window. The caveat is that you need a bit of knowledge of Comfy UI + Custom Nodes Manager, but it works perfectly. After all, in Ambrosinus, Grasshopper is used only as a graphical interface for interacting with SD models.

[Ambrosinus (Luciano Ambrosini)]

Firstly, thank you, @VAT , for your interest in ATk! This means a lot to me as I am not part of a software house, but an independent researcher collaborating with other researchers and universities.

Having clarified this, I apologize for the delay in responding on the forum, which was due to some urgent work that needed to be expedited in July/August. I also invite anyone interested to check out the latest updates on my website.

Regarding your .gh file, to prevent various errors,

I always suggest starting with the demo file available on my GitHub page (the latest version). I tried your prompt on an outdated demo file, which resulted in a 500 server error, but it worked correctly on the v129 version with very simple settings.

I encourage you to test using the demo file I referenced. (as a reference image I used the Revit Simple Project House )

On another note, the A1111 project is regularly updated, and I will revise the ATk version as soon as I receive the notification. Currently, I haven’t updated some procedures that are flagged in the Stable Diffusion log as deprecated and automatically corrected – I plan to address these updates in the coming days.

Finally, a response to @Chris_C 's message.
I understand the frustration of trying to use research projects or plugins that are frequently updated (so that’s why I appreciated your interest in this too). I want to emphasize that research projects, especially ATk, were created as free projects accessible to everyone (indeed, many design teams use it, and others occasionally consult it to incorporate it into their workflows for small tasks).
However, I realize that if you’re seeking a high-performance product from a hardware perspective (your own aside), like ComfyUI and the use of Google Colab for instance, you certainly have the option to use a platform that emulates Grasshopper and will likely produce highly controlled images. If this is the primary goal (rather than exploring an interaction between GH + SD + User), then I understand your need for semi-open products, or I would suggest considering paid solutions that deliver remarkable results (now even Microsoft Copilot and Claude are gearing up, not to mention Grok, Flux, and Leonardo…).

Since its inception nearly two years ago, ATK has been free, and if you have a good GPU, you can enjoy all the small pleasures it offers.

I always thank you all for your interest and, in particular, for the constructive feedback that greatly assists the project’s growth and reminds me that it all stems from a passion fueled by enthusiasm!
Thank you & Stay Tuned!