MAIN INFO Ambrosinus-Toolkit latest version is: 1.3.3 - CHANGELOG v1.3.3 -1 The "LaunchSD_local" component has been updated, now it receives input data from the new "SDModelPaths" component; -2 Added new "SDModelPaths" component, now the user can specify the full path of specific system folders for checkpoint, VAE, hypernetworks, LoRA and embeddings models; -3 The Atk_MultiCN component has been updated with the launch of the "SDProgressMonitor" window; -4 Integrated new component "Gemini ImageGen" that uses Google AI (Gemini Flash Preview"). This component requires the insertion of a Google API key that can be generated for free. It is likely that due to recent geopolitical restrictions, it is necessary to use a VPN to run the component for free from your geographic location; -5 The "launchSDopt" component has been replaced by two new components, "LaunchSDchkp" and "LaunchSDvae". Implementation due to the integration of "SDModelPaths" and therefore the possibility of indicating personal paths to the folders of the processing models downloaded locally by the user; -6 Introduced a new sub-category "Geom", which will collect new components dedicated to geometric tasks; -7 Added new component "MinBBox", minimum bounding box; -8 Added new component "MinBSph", minimum bounding sphere; -9 Some new features were presented at the Envisioning Design Day event held at the Instituto Europeo di Design (IED) in Milan on June 14, 2025, the video speeches will be published in the next few weeks; v1.3.2 -1 T2I (single ControlNET unit = sCNu) - "AItxtToimg" component has been updated with two sets of features in the context menu. Added ControlNET general settings with "low vram" and "pixel perfect" options. Added Sampling type with Automatic, Karras and Uniform sampling system. Fixed and enriched the infotext layout embedded to PNG generated image file; -2 I2I (sCNu) - "AIimgToimg" component has been updated with three sets of features in the context menu. Added ControlNet general settings and Sampling type as AItxtToimg component, in addition now there is a WebUI Mask settings where user can set up what type of masking behaviour SD must run such as inpaint not masked option (by default it will inpaint the masked one), Fill/Latent Noise/Latent nothing behaviours. Added the Mask blur settings to improve the tranbsition area between the masked area overlapped to the original source image. Added the IP2P processor for the inpaint-process. Fixed and enriched the infotext layout embedded to PNG generated image file; -3 Outpainting (sCNu) - "AIoutpaint" component has been updated with two sets of features in the context menu. Added ControlNet general settings and Sampling type as AItxtToimg component. Fixed and enriched the infotext layout embedded to PNG generated image file; -4 Added a new component named "ARatioImg". This component helps set image size, width/height dimensions and the image Aspect Ratio. I provided a bunch of pre-setted AR to speed up the choice. In the context menu, it is possible to assign a different component's behaviour to assign the maximum size in pixels to the minimum image's dimension (by default the component assigns MaxPx to the maximum one according to the image's AR); -5 "OpenDir" component has been improved in performance also in providing the filename and the full path of the input image, which is very helpful in setting up AI component; -6 "ImageMask" component as been improved in managing the final output image. Now it is possible inver colors from b/w to w/b and even choice the output format when the user assigns a back/mask custom colors; -7 Added a new component named "SetPreProcessor". This component can generate and partially set mask images elaborated from ControlNET processors (canny, depth, lineart etc.). These masks will be generated in a temporary folder so when the user switches the component to false that folder disappears, in any case it can turn off this behaviour by setting true the option in the context menu, in this way the preprocessor mask image can be used even in a simple I2I process as simple mask; -8 Fixed the "SetCanvas" component; -9 According to the above updates I have improved, fixed and updated also the Windows OS App "WinSDlauncher" to version 1.1.4. This App can be used side by side with ATk plugin for Grasshopper especially to run this latter over your LAN. I recommend reading more about it on my website/GitHub page; -10 Fixed seed input params to manage 10 digits format; -11 Updated and fixed the "Upscaling" component; -12 Added a new component named "SDstyles" for the T2I process. This component will read and apply 250+ styles available in a CSV file provided and shared on my website/GitHub page, please download and install it if you want to use this new component with "AItxtToimg". In the context menu you can define two different behaviours, show styles as numbers or with their names; -13 Improved significantly the "ViewCapture" component. This component can now interact with the Rhino viewport, allowing the user to easily set the aspect ratio of the screen-captured image. This means that you can save your 3D model image efficiently and have it quickly ready as an input image in the T2I process. Within its context menu, the user can set up the real-time aspect ratio vision overlapped to the Rhino canvas and even adjust the aspect ratio according to the viewport. According to ARatioImg, the user can select from different aspect ratio presets (10+) to quickly adapt the best aspect ratio to her/his 3D model; -14 According to the possibility of running ATk in LAN mode, each GenAI components (T2I/I2I/Outpaint) when run activate and launch the "SDProgressMinitor" window. From this window, the user can follow and manage the StableDiffusion generative progression, skipping and even interrupting it without forcing the shutdown of the CMD window (WebUI client black window). this feature is fundamental to follow the advancement during the LAN running; -15 Added and Updated the ATk installation user guide downloadable from the Ambrosinus Toolkit official page reachable from all AI components by clicking on "AmbrosinusDEV log" from their context menu; -16 "AtkImageViewer" component has been updated and fixed; -17 "GrayGaussMask" component has been completely coded into C# language. Now it is available within ATk and anymore as a Python cluster; -18 The Multi ControlNET (mCNu) are currently in development (it outperforms the capability to transfer styles to different images). I have already demonstrated it on my LinkedIn profile, but for this v1.3.2, I opted to postpone its release to the next ATk version due to the time-consuming process of updating and developing this toolkit completely for passion (and improving users' workflow); v1.3.1 -1 I have updated ATK to version 1.3.1 to fix an exception that was causing a "loop execution". v1.3.0 -1 I have updated ATk with the latest Stable Diffusion Annotators, Preprocess model and Samplers; -2 Integrated the ControlNET mode to give priority to prompt/CN/Balanced; -3 Improved the metadata info text output embedded into generated images. Now it shows WebUI and ControlNET (A1111) versions; -4 Added the "ATkImagePreview" component into the Image subcategory. This component displays images on the Grasshopper canvas and can be scaled by a factor; v1.2.9 -1 I have updated some minor bugs encountered when running "canny" mode through the "AItxtToimg" component; -2 Added auto-populate feature to all value list components involved in the grasshopper definition and AI Stable Diffusion components. Now ControlNET lists (samplers, resizing features and checkpoints annotators) will be updated automatically simply by connecting the default value list Grasshopper component; -3 Added the "ValueListPopulator" component into the Manage subcategory. Through this component, you can set a custom value list by passing your keys and items. It is possible to add quotes as well; -4 Added a funny "ATk Support BuyMeACoffee" component into the About subcategory just to share a moment of virtual friendship (if you wish); -5 As always Stay tuned! v1.2.8 -1 Added new component "AIoutpaint". Now You can explore everything outside of the image frame; v1.2.7 -1 This version (v1.2.7) has been updated to the Grasshopper core 7.35.23346.11001 so user can get the toolkit inside Rhino v7 and Rhino v8 (all C# components); -2 Minor bugs fixed; -3 Added "SetCanvas" component (light version). You can customize partially your canvas and components; v1.2.6 -1 Added AI variation mode locally into the AI subcategory. Now You can run Stable Diffusion with this feature you can alter the images through the "Denoise" parameter and the use of the editing mask; -2 Added Text-to-3D, Image-to-3D and text-to-texture from mesh.ai service into the AI subcategory. The "MeshyReq" generates the request while "MeshyRet" gets all files elaborate through the API key (WIP in the next few days, I will upload the "ghuser" objects to the GitHub page; -3 Added the "B64toBMP" component into the Image subcategory. Through it is possible to convert an image to Base64 string (text) and this latter into a Bitmap. It is useful to embed an image into a gh file without attaching the bitmap file; -4 Added the "Alkemyc Kaballah" component into the Extra subcategory. It is a funny experiment in numerology according to the Kaballah tradition - "What's in the name?"; v1.2.5 -1 I have fixed different bugs; -2 Updated "GradientGen" and "GradientDecod" to generate multiple Gradient components and decode a multiple list of RGB values; -3 Added "SaveGradientLib" component into the Manage subcategory. Through it, you can save different Grasshopper gradients in a custom library format (grasshopper_gradients.xml file); -4 Added "SwitchGradientLib" component into the Manage subcategory. Now it is easy to switch among different gradient libraries without rebooting Grasshopper; -5 Added "WireX" component into the Manage subcategory. Say goodbye to tedious connect/disconnect action with multiple components. Now you can control individual input/output parameters when connecting and disconnecting components; v1.2.4 -1 I have fixed different bugs that come up after some A1111 API updates. The issue generally appeared with this error message: "Index was outside the bounds of the array". it was related to the info text embedded in the generated image. -2 Added an internal timeout in the "SDopts_loc" component (that one to select different checkpoint models); -3 Fixed an issue related to the possibility of generating more than one image (N params); -4 Fixed different issues about the INFOtext generation if the user switches among T2I Base, CN v1.0 and CN v1.1.X processes; -5 Fixed the possibility to assign CFG decimal values (e.g.:5.5, 5.8, etc.); -6 Reduced the number of times that the CN v1.1.X alert message turns on; v1.2.3 -1 I have fixed a bug that comes up after some A1111 API updates. The issue generally appeared with this error message: "Index was outside the bounds of the array". it was related to the info text embedded in the generated image. v1.2.2 -1 "LaunchSD_loc" has been updated: now the input parameters give the progression sense of the sequence of the steps. Now you can check the IP address through the integrated ipconfig command helpful when the user ticks the listen argument; -2 "CustIPport_loc" component has been added: this component helps the user in setting the Remote PC configuration passing the IP address shared on his network and the port number opened on the Windows OS Firewall; -3 "AIeNG_loc" component: its new look added the implementation of the ControlNET version 1.1.X (bear in mind it is implemented, but not yet fully API supported so I suggest using v1.0) so you can test it but it is already known that the dataset-models do not work properly in the API mode; -4 Thanks to WinSDlauncher (the Windows OS app) is it possible to set the Server PC without switching/fetching the Rhino license between the Server PC and the Remote one (please see the video demo); v1.2.1 -1 "LaunchSD_loc" has been improved, now it is possible to interact with webui-user.bat file and some of its most used arguments. It can create webui-user.bat file according to cmd arguments; -2 "ViewCapture" component has been added to the 2.Image sub-category. This component allows the user to save the Rhino viewport as an image file (JPG/PNG) with a scale factor and as a Rhino Named Views object. Now you can pass easily your 3D model screenshot as BaseIMH input in the AIeng_loc component to render your concept design; -3 "UpsclAI_loc" component has been added to the 3.AI sub-category. This upscaler can improve the output size of any image format leveraging AI models working with the A1111 project; -4 "OpenDir" component has been added to the 4. Manage sub-category. It is still in WIP mode but basically, it can open quickly any full folder path passed as input; -5 "AIeng_loc" component has been updated. Currently, it works fine with ControlNET v1.0 extension (see the video tutorial on how to install it) but it is ready for ControlNET v1.1, more info in the next update; -6 "Extra components" have been re-ordered in the 9.Extra sub-category. Please, delete old ones (to avoid an overlapped effect) and download new versions from the GitHub page; v1.2.0 -1 "SDopts_loc" has been added to the AI sub-category. This component allows the user to set a custom Stable Diffusion model checkpoint, for instance, using the “mdjrny-v4.safetensors” model, a dataset trained with Midjourney version 4 images; -2 "SD-Imginfo" allows the user to read all AI-Gen settings used for image generation; -3 All C# components have been updated with a new right-click context menu to get more info; -4 AI subcategory components disposition has been reordered; v1.1.9 -1 "AIeNG_loc" has been added to the AI sub-category. This component lays the foundations for future adaptations, expansions and tools to make the most of the best features provided by the AUTOMATIC1111 project - It is part of the "_loc" components able to run Stable Diffusion and ControlNET neural network locally; -2 "LaunchSD_loc" has been added to the AI sub-category. This component can open and close the localhost port and also check the status - It is part of the "_loc" components able to run Stable Diffusion and ControlNET neural network locally (thanks to the Automatic1111 project); -3 "LA_DPTto3D_b101" has been added to the AI sub-category (downloadable from the GitHub page). This component can generate a PointCloud (stored in the PLY file format) directly from a 2D RGB image. It exploits the DPT technology (transformers libraries). Right-click on the component for more info; v1.1.8 -1 "FileNamer" component has been fixed; these updates (v1.1.7/v1.1.8) were necessary due to a couple of incoming research projects; -2 "La_GrayGaussMask" has been updated. Some minor fixes; -3 "LA_OpenAI-GH_Ask_b107" has been updated. Some minor fixes; -4 "LA_OpenAI-GHadv_b112" has been updated. Some minor fixes; -5 "LA_StabilityAI-GH_b108" has been updated. Some minor fixes; v1.1.7 -1 "SeeOut" component has been fixed to avoid a loop-action if its output is passed as input in OpenAI or StabilityAI component. Now it is not necessary to rename the Slider Number with the name "SeqID"; v1.1.6 -1 Ambrosinus-Toolkit updated with the "SeeOut" component in the Image sub-category. It can slide the image file path Output by a "SeqID" slider or just see the latest file path generated from OpenAI or StabilityAI GH component; -2 Minor fix for LA_StabilityAI-GH component, now at Build-107, see the GitHub page to download it. If you already downloaded this build on 2023/02/06, please do it again. -3 Minor fix for LA_OpenAI-GH component, now at Build-111; see the GitHub page to download it; -4 Minor fix for LA_OpenAI_Ask-GH component, now at Build-106; see the GitHub page to download it; v1.1.5 -1 Ambrosinus-Toolkit updated with "FileNamer" component. It can generate a filename to avoid overwriting action; -2 New version of the "SdINinfo" component. Now able to show different info from StabilityAI image output like BaseIMG and MaskIMG info/links; -3 "LA_StabilityAI-GH" now is advanced, which means that is possible to run: TXT2IMG, IMG2IMG and IMG2IMG Masking features. I added CLIP guidance too in order to drive the image generation process; -4 All images will be stored in IMGs folder. All metadata will be stored in TXTs folder. Each run will be tracked in a unique CSV file. Helpful for data storytelling; v1.1.4 -1 "LA_StabilityAI-GH" build-102 has been updated. After the StabilityAI updates of the "stability-sdk" Python library to v0.3.0, this component you can now select different Engine (Stable Diffusion v1, v1.5, v2.0 and v2.1) and among 10 different samplers; I have modified the filename layout for integrating "SD-INinfo" actions; -2 "SD-INinfo" component has been developed and added. This component allows the user to grab settings info from the filename of the image generated by LA_StabilityAI-GH component; -3 fixed minor bugs and inaccuracies in descriptions and names; v1.1.3 -1 "AnsToPrompt" has been added. This component converts the AskToOpenAI answer into a text prompt; -2 "AskToOpenAI" component has been developed and added (see GitHub page to install it) to play with OpenAI completion mode, the OpenAI GPT-3 model transforms this component into a super smart "chatBot". I asked for design tips but also for generating a simple piece of code (Python, C# etc.); v1.1.2 -1 Subcategory Layout has changed; -1 "DALLEfromGH" UI updated with "Light version" info; -2 "LA_OpenAI_GHadv" has replaced the CPython "Light version". Now you can experiment with the EDIT and VARIATION modes. You can install it via the GitHub repo; -3 Subcategory Image has been added; -4 "ImageConv" can read image file info and can convert images to these formats: Bmp, Emf, Exif, Gif, Icon, Jpeg, MemoryBmp, Png, Tiff, Wmf; -5 "ImageMask" can generate PNG and JPG image masks by simply drawing them inside Rhino over the BaseIMG; v1.1.1 -1 "DALLEfromGH" has been added to the toolkit. In order to run it the Toolkit uses the Json DLL library provided in the installation ZIP file. Now you can explore OpenAI (DALL-E) prompt-to-image requesting process directly inside GH. No Python libraries need to be installed to run OpenAI. It has replaced the LA_OpenAI-GH.ghuser component (anyway it will always be available on GitHub and you can always install it by following the guide. v1.1.0 -1 "GradientGen" has been replaced by Ambrosinus-Toolkit project; -2 "ToolkitVersion" component - Now the toolkit can show the version and the main changelog (main updates) and above all notify user if he needs to install the latest version; -3 "HEXtoRGB" now can be covert also from RGB to HEX values with lowercase and hashtag options; -4 "KelvinToRGB" converts a Kelvin temperature to RGB value (for specific values shows some extra info about devices that emit the same temperature); -5 "WavelengthToRGB" converts a Wavelength in a visible light range (380nm-780nm) to RGB value; v1.0.0 -1 "GradientGen and Utilities", an internal module part of Ambrosinus-Toolkit project;