ExcaliAI Enhanced: More Visual Thinking Power  ## Metadata - Author: [[Zsolt's Visual Personal Knowledge Management]] - Full Title: ExcaliAI Enhanced: More Visual Thinking Power - Category: #articles - Document Note: --- tags: [#Obsidian, #ExcaliAI, #VisualThinking, #AIEnhanced] - #Obsidian - #ExcaliAI - #VisualThinking Author: Jolt (YouTube Creator) Source: YouTube Video – *ExcaliAI Enhanced: More Visual Thinking Power* Reference: [Transcript & Summary](https://www.youtube.com/watch?v=3G8hsV-V-gQ) date created: 2025-08-23 23:59 --- # Title: ExcaliAI Enhanced: Improved Image Generation and Use Cases ## Table of Contents: - [📝 Summary](#-summary) - [📖 Main Content](#-main-content) - [🗒️ Notes](#️-notes) - [📖 References](#-references) - [🗣️ Paraphrase](#-paraphrase) - [📰 Newsletter](#-newsletter) - [✅ Action Items](#-action-items) - [🧪 Hypotheses](#-hypotheses) - [📊 Methodology](#-methodology) - [❓ Questions](#-questions) - [🔮 Future Research](#-future-research) - [💡 Implications](#-implications) - [🧠 Learning Styles Assessment](#-learning-styles-assessment) - [📊 Accessibility Assessment](#-accessibility-assessment) - [🌍 DEIB Assessment](#-deib-diversity-equity-inclusion-and-belonging-assessment) - [✅ Logical Validity](#-logical-validity) - [🔍 DARVO Analysis](#-darvo-analysis) - [🔍 SEO Evaluation](#-seo-evaluation) - [🎯 Conclusion](#-conclusion) --- ## 📝 Summary: The video introduces **ExcaliAI Enhanced**, a script for Obsidian’s Excalidraw tool, which leverages OpenAI’s models for generating and editing images, visual brainstorming, and even code prototyping. Key upgrades include: - Diverse style generation from sketches (sci-fi, fairy tale, minimalism, etc.). - Editing capabilities with masks. - Automated quote-to-image visualization. - Mermaid graph generation and editing. - Interactive use cases like calculators and web apps. The tool requires an OpenAI account and API key setup but provides significant creative and workflow value for a modest cost. --- ## 📖 Main Content: ExcaliAI Enhanced extends Excalidraw beyond static drawing, integrating **AI-powered workflows** for visual thinking. By combining **image generation, editing, brainstorming, and structured diagramming**, the tool transforms how users capture and expand ideas. - It bridges raw sketches with polished illustrations. - It turns prompts and visual elements into **dynamic outputs** (images, graphs, apps). - It maintains flexibility: users can adjust prompts, system settings, and graphical orientation. The demo highlights real-time use cases: creating horror images from pencil sketches, editing busy cityscapes into beaches, automating inspirational quote illustrations, and prototyping working calculators. The result is a **multi-modal augmentation of creativity**, where Excalidraw becomes a sandbox for design, reflection, and innovation. --- ### 🗒️ Notes - [00:04](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=4) Overview of enhancements, capabilities, and installation. - [02:54](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=174) Image → prompt → style variations via OpenAI Vision + DALL-E. - [05:47](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=347) Editing images: city vs. beach. Masking explained. - [08:40](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=520) Automating quote illustrations, UI walkthrough. - [11:25](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=685) Symbolic landscape image contrasting past and future. - [14:23](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=863) Mermaid graph editing: structure, orientation, arrows. - [17:10](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1030) Functional calculator generated, respects color scheme. - [19:46](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1186) Experimental shape generation (smiley success, circle fail). - [22:22](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1342) Visual brainstorming produces associative imagery. - [25:01](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1501) Under-the-hood: element selection, task choice, system prompts. - [27:37](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1657) Setup: API key, funding, script installation. - [30:04](https://www.youtube.com/watch?v=3G8hsV-V-gQ&t=1804) Conclusion: costs low, benefits high. --- ## 📖 References: - Jolt, *ExcaliAI Enhanced: More Visual Thinking Power* (YouTube) - OpenAI API documentation - Obsidian Excalidraw plug-in docs --- ## 🗣️ Paraphrase: 1. Original: *“When the image is generated, OpenAI deletes the URL after 30 minutes.”* Paraphrase: The generated image link is temporary and must be saved locally within half an hour, or it vanishes. 2. Original: *“ExcaliAI provides mind maps that partly summarize, partly challenge the ideas.”* Paraphrase: The tool doesn’t just outline concepts—it also critiques and expands them, encouraging reflective thinking. --- ## 📰 Newsletter: 🎨 **Visual Thinking Goes AI-Enhanced in Obsidian** ExcaliAI Enhanced takes **Excalidraw sketches to the next level** with AI-driven creativity. From transforming doodles into polished sci-fi art to generating quote-inspired illustrations, the script integrates OpenAI’s Vision and DALL-E into your Obsidian workflow. Use cases span from **visual brainstorming** and **Mermaid graph editing** to even **functional web calculators**. The setup is straightforward—install the script, plug in your API key, and watch your notes evolve into interactive canvases. 💡 If you’ve ever wanted your ideas to literally sketch themselves into new forms, ExcaliAI is worth the few cents per image. Try it here: [ExcaliAI Enhanced](https://www.youtube.com/watch?v=3G8hsV-V-gQ) --- ## ✅ Action Items: - Frame ExcaliAI as a **case study in contributor education**: show how OSS docs could integrate visuals + AI. - Connect workflow to **WordPress training handbooks** for onboarding contributors visually. - Explore AI image-gen for **event recaps and dashboards** in The WP Community Collective. - Advocate for **AI-enhanced visual literacy** as part of OSPO governance: making diagrams + mind maps more accessible. --- ## 🧪 Hypotheses: 1. AI-augmented visuals increase comprehension and retention in OSS contributor docs. 2. Automated quote-to-image workflows improve community engagement in training content. 3. Mask-based editing enables more intuitive interaction with AI tools. ### 📊 Methodology: - Controlled experiments: test contributor recall with/without ExcaliAI visuals. - Usage logs: analyze which use cases (graphs, brainstorming, apps) dominate adoption. - Surveys: gather perceptions on clarity + creativity. ### ❓ Questions: - How can AI-generated visuals support multilingual contributors? - Could ExcaliAI diagrams integrate into WordPress Gutenberg natively? - What ethical guardrails are needed to prevent misleading auto-generated visuals? ### 🔮 Future Research: - Linking ExcaliAI outputs to **contributor dashboards** (e.g., CHAOSS metrics). - Exploring AI support for **design accessibility** in OSS projects. - Testing whether visual brainstorming sparks **new contributor ideas**. ### 💡 Implications: - Strong potential to **democratize documentation** by lowering visual design barriers. - Could influence **open source governance UX** by standardizing AI diagrams. - May reshape contributor training into **multi-modal learning experiences**. --- ## 🧠 Learning Styles Assessment: - **Visual**: Excellent (images, graphs, mind maps). - **Aural**: Missing (no audio summaries). Suggest podcasts or TTS exports. - **Verbal**: Strong (text prompts, descriptions). - **Kinesthetic**: Moderate (interactive editing workflows). - **Logical**: High (structured diagrams, calculators). - **Social**: Low (individual focus). Suggest community collaboration. - **Solitary**: Strong (self-reflection features). - **Naturalistic**: Absent. Could use analogies (ecosystems, environments). --- ## 📊 Accessibility Assessment: - Images lack alt-text → barrier for screen readers. - Strong keyboard use (mask toggling, hotkeys). - Contrast/readability depends on Obsidian theme; not standardized. - Image persistence requires manual save—could hinder neurodiverse users. --- ## 🌍 DEIB Assessment: - **Cultural relevance**: Narrow, tech-cent - Summary: ExcaliAI Enhanced: More Visual Thinking Power discusses the various enhancements made to the ExcaliAI script over the past few weeks. The script is available in the Obsidian Excali Draw Script Store and is used for generating images and illustrations based on prompts. The article provides examples of different use cases, such as generating images in various styles, editing images with masks, and automating the process of creating images for daily quotes. It also explores features like generating mind maps and code generation based on user prompts. The article highlights the capabilities of ExcaliAI and provides instructions on installation and setup. - URL: https://youtube.com/watch?v=3G8hsV-V-gQ&si=pEVi8h7LZhYW0xGx hey everyone jolt here exol AI is a script available in the obsidian exol draw script store and today I'd like to show you the different enhancements I've made to the script over the last couple of weeks first of all we're going to look at a number of examples to Showcase what exol AI is capable of then we're going to take a bit more structured look at the various use cases currently in implemented we're going to look under the hood just a little so you have an understanding of the logic flow of how to use exol Ai and finally we're going to look at how you can install it how you can set up your API key and how you can get started with excal AI so let's start with this example so I have this drawing here I'm going to start exol AI right here so it's in the obsidian tools panel and if I scroll down here you will find the robot here that's xol AI because we are going to use this frequently today I'm going to long press on this and by long pressing on it it was added here to the side panel so it's easier to access and now I'm going to click on this and here for the first example I'm going to choose the model that I want to generate an image from from an image and a prompt and here I'm going to just add horror I think that's enough so I want to create a horror image of this here I can set the size I'm actually going to set this to be portrait and here's an important notice that I'm going to explain as this generates note the generation takes quite some time because this depends deps on open AI response times the note I wanted to make is when the image is generated you receive the image with a URL from open AI that URL is going to be deleted after 30 minutes so if you want to keep the image then you will need to do the following you will need to right click on the image and you need to choose save image from your L to local file when I do this then the image was saved locally now take a look at what was generated I think this is pretty cool so first of all you can actually see the tree you can see the house with the Red Roof you can see the road you can see the flowers or vs in this case and indeed this is a horror image now what happens under the hood is I send this image to the vision interface of open Ai and ask it to generate a detailed prompt and so when the image is generated this is the detailed prompt that openai created which was then sent to doll e for the image generation and this is how we ended up with the image now let's just take a look at some of the examples we have so I'm going to pull this here and I'm going to put my horror image here and let's just quickly look at the various styles that I created so this was my sci-fi image this is my fairy tale version this is my mountain cabin version this is my simple life version vangog minimalistic pencil version the photo unsplash and the black and white photo version I think these are all pretty cool and of course in all of the cases you can see how the house the original pencil drawing that I created in exol draw on my phone was translated here now let's look at just another example here so this is again an example here of a city and you can see here these were the various versions that were generated from this I'm not going to spend the time now to generate them but just to show you how this works now what I want to show you is how you can edit an image so this is my image edit use case right here so you can see that I can add a mask over the picture now notice that the generated image is a square and that is a limitation by open AI dolly that the image must be squar shaped but here I can maybe change now my prompt I'm going to delete or just move my mask from there and I'm going to use these two masks on the side and I'm going to now select the image I'm going to start excal Ai and I need to choose a different prompt I'm now going to choose edit image now notice that when I click edit image and maybe let me just click this back so you can see it now you can see the image with the gray boxes around it if I click edit image then you will see that this suddenly turns into two now it this one right now maybe is not the ideal one let me just put this back here for a second so you can see what's going on I'm going to again click edit image and you can see here that the black areas are where the image is going to be created this one right here is the same size as this one it's just a bit because they are next to each other you don't see the mask but of course you can see the mask right here and what I want to do to experiment with is let's see what if I say that this is a busy City street so this might not actually work so well let's see what open AI doly is able to do this with this because of course this image is not a busy City street but maybe it is going to draw some houses on the side or something like that I'm super curious how open AI is going to solve this problem for us so let's just wait until this generates and well well yeah so this is not really perfect so you can see the busy City street this thing appeared here so my earlier approach with the uh space ship Flying Saucer was better but maybe we can do just one more experiment you can see here I generated ports here so I'm going to take this image right here I'm again going to click exol AI and I'm going to write here City Beach with lots of people so hopefully this is going to work out much better note the generation can take up to 20 30 seconds and wow this is so cool so you can see here our beach was created based on this image at the top just a continuation of the image I'm actually going to save this so let's now look at one more use case the use case I want to look at is generating a prompt by the way notice that every time I generate an image this message pops up just to remind you that if you want to keep the image you need to save it I'm going to close it now I have this practice of creating a picture every day for a daily quote so this is a quote that I created this image for with mid Journey so this is not excit and not Dolly this was my image generated for this quot to at Mid journey and I usually spend good 10 20 minutes playing around until I get to an image that I really like but right now let's see what happens if we automate this process so I'm going to start exol AI I'm going to choose the generate an image to illustrate a quote and I'm going to just paste here the quote now in the meantime let me just give you a little bit of explanation first of all let me close this because I shouldn't have selected the text I don't want to send the image so now it's only going to be the user prompt so let me just walk you through what you see here on the screen we have already looked at this drop- down list this is where you can select the action each of the actions has a short help so if you read it hopefully that's going to give you some better idea of what that prompt is about I have a system prompt for most of the actions you can read the system prompt here you can also try to modify the system prompt I played a bit with these system prompts so they generate the type of result I want you can move modify this as well if you want and see if you can come up with a better system prompt and of course here we have the user prompt and here I can choose the size of the image I would like to see generated so this is now going to be a portrait image note that if you configure Dolly 2 we'll look at configuration later then the image sizes are going to be different but right now let's just generate an image to illustrate this quote and let's see how this image is going to look like now this is also something that happens sometimes that there's an unexpected error no problem we can run again and usually the second time when you run the script it executes in the script I'm not doing serious amount of airor handling and as a result you sometimes get that back but again this is just a reminder that as long as you see this generating spinner here you know that I am waiting for the response from open AI because I print the image prompt here this usually gives you an understanding of the logic why open AI thought that this would be a good illustration for the quote so again our quote was don't ruin the present with the ruined past so let's read this create an image featuring an otherworldly collidoscope like landscape the foreground includes a figure standing at a Crossroads their back to the viewer exuding resilience to their left a path showing the past cluttered with fragments of broken objects and Faded memories to their right the path is bright leading to a vibrant and Lively landscape the figure 's outstretched arm reaches for the enriching future Rich contrasting colors should exude energy and optimism the scene should blur reality and fantasy boundaries using light and Shadow to symbolize a journey from past Darkness to present light the overall feeling should be hopeful capturing the promise of embracing the present so I think with this explanation this image is pretty good the next one is going to be about creating a mind map to explain a whiteboard image this was available in the earlier version of exol AI already I made a little change that I want to show you here you need to choose challenge my thinking right there and challenge my thinking is going to generate a mind map you can add a user prompt or you can leave the prompt empty going to submit this this is going to generate a mind map that partly summarizes the ideas on the image partly adds additional ideas to it and partly challenges the idea so the image was or the Mind map was generated but wasn't inserted so there's probably an issue with the script and here's what's new so first of all I inserted the mermaid script here and secondly here if you open the menu and go to mermaid to exol draw you will find the script here and you can try to manually fix it unfortunately open AI CH GPT sometimes results in incorrect formatting of mermaid diagram so I'm going to just now click clean this up probably this should be subgraph or something like that but let's look at how this looks like I'm guessing that if I would type here so let's just undo the previous changes so let's just type in subgraph like this and I'm going to type end and I'm going to just put this on my clipboard and replace these so this unfortunately happens sometimes that ch gbt generation of a mermaid graph is not perfect but the good thing is that you don't need to regenerated of course you could also just simply press generate and create a new cph or you can come in here and you can edit it the additional benefit of editing the graph is and so there you go it's now created this graph that I can also change this to be from instead of top down maybe to be left right organized and now if I insert this right here then I can also use my script select elements of type and I'm going to select the arrows and here I can make the arrows to be sharp instead of rounded I like this look of the Mind map a little better and now we can read what open AI returned to also how should we go about filling the page and this is the brick road approach to sketch noting and this is the expansion of the concept so this is an alternative to linear note taking and these two lines on top of are on top of each other so I'm just going to bring this down here and yeah I'm not going to read through the material but you can see here that there's this analysis of my drawing I can use this as a reflection on my thinking I'm going to do this after recording this video but I thought that this is a pretty interesting use case the other example I want to show you is the code generation which I think is more of just a game I'm going to use this calculator right here and I'm going to copy the prompt from here and now I'm going to again start exol AI I'm going to select here wireframe to code the new feature in this case is you can provide the description of the application you want to build as a user prompt so it's not part of the image so you could also add the text prompt under the image and send it as an image and then open AI would decode the text and generate probably the same calculator this gives you some flexibility and options that now you can add a separate prompt so let's look at this calculator it looks good so let's just see 12 plus let's maybe 5 \* 6 equal 42 so the calculator actually works as expected 2 over one is 0.5 so this is this is pretty good I'm I'm pretty happy with the calculator also in the description I asked the calculator to follow the color scheme of the image and indeed I can see that this is largely the color scheme I'm happy with the outcome and now before we quickly recap the various functions and look under the hood how to use the tool here's one last example so in this case I took a photo from the internet and I added some icons from Flat icons here and as C GPT or rather Dolly to create an image so here's the resulting image I think this is uh pretty nice and it's a nice mix of using some icons from the internet as well as a picture to frame what I'm looking for and then GPT can generate the image for us so with this let's now just recap the different use cases so alt together there are three big areas of use cases well you can play and that's the wireframe to code web development use case you can create illustrations and this reminds me there's one more example we didn't look at so you can also create excital elements and convert elements to shapes so let's look at this I'm going to draw this circle here I'm going to select this I'm going to select convert sketch to shapes I'm going to delete my prompt and let's run this this should be somewhat faster because here there's no image Generation Um it was faster but the end result is well not very good so this was converted into a box I mean I thought that this was a circle um so it's it's not perfect let's look at the other use case these two use cases about creating EXC colid draw shapes is more of an experiment from my side they are really not that good at the moment but let's create a smiley and let's see if the smiley is created there's our Smiley if I can get the smiley out of there so the smiley is actually pretty nice I think that's a I'm I'm happy with the smiley part the they can also try others I was successful with generating a car and a table and small things like that but again sometimes what's generated you need to remove background colors or make Parts transparent because it's hidden it takes a bit of experimentation to work it out but back to a look at the use cases so web development and play the other is generating EXC colit Elements which is super experimental creating and editing images we looked at a whole bunch of examples here um there are also additional use cases where you can just simply generate an image from a prompt you can also use this for visual brainstorming maybe we can look at this example as well so back to the brick road approach the visual brainstorming is about creating an image so I'm going to now choose visual brainstorm I'm just going to send this image over and I'm going to click run so the visual brainstorm creates a picture that somehow resonates with the content of the Whiteboard that you've created my idea is that looking look at the picture and looking at the Whiteboard it can spark some additional ideas so this is the image that was generated so this is about sketch noting for sure we have the sketch note artist we have this idea up here we have a bit of a diagram up in this corner it's actually this Leaf of ideas that are spreading all over it's it is a nice illustration is it helping me thinking about sketch noting well maybe it does so I can reflect on this image and connect it with the ideas from The Brick Road approach to skep thing and just see some links I'm actually going to save this image so again final reminder that you need to save the image locally if you want to keep it otherwise you come back tomorrow and this image is no longer going to be available I think we've looked at all the use cases so the other use case apart from illustrations is to generate new ideas and generate new ideas could be illustration of a quote it could be a visual brainstorm or a mermaid graph that illustrates an idea so then let's look at what's under the hood I'm going to be very quick on this but I think this is helpful and you also saw how I'm using this in the presentation so you need to select the elements you want to include in your prompt on the canvas so for example like this I can select these elements and then these are going to be the ones that I'm sending to open AI for some of the prompts like for example for the quote you don't need to select elements because then you're only sending the text prompt and that's okay then you choose the task now when you choose the task then depending on the type of task you saw that in case of selecting the image edit then the picture and the excolo elements were taken in two separate parts so you remember here that we were looking at this mask like that and here if I start excal AI you can see that in the normal case this entire image would be sent to open AI if I choose here edit image then this is split into two I have the picture and I have the mask and wherever the mask is black that's where the image is going to be generated maybe this can be much more visible if I do it here let me bring this to the front like this and so for example if I would want to change this house into a skyscraper and then I would send I would open exol AI like this and then you can see here that this is where the skyscraper will go you can see the mask and you can see the image so depending on the task you will see a different preview image either the image plus the mask or the image only you can then provide your text prompt and optionally modify the system prompt so the system prompt is this one up here in case of edit image there is actually no system prompt but most of the cases there are so this is the customize system prompt part and then when you hit run at the bottom then depending on the activity you selected different things will happen either a mermaid graph is generated or an EXC colro drawing is generated or a web app or maybe dolly is executed and an image is created in some cases and this is the case then you create an an image prompt and you want you add an image and a prompt and you want to create a new image based on that inspired by that like my little house so this was uh these are all the examples of that then what happens is first I call GPT vision and GPT Vision generates an image prompt and then with that I go to Dolly and I generate the image and finally if you like the image you should save it locally right click save locally or if not then open II will delete the image in 30 minutes I'm not saving the images automatically to the Vault because I'm assuming that there's going to be lots of rubbish generated by open Ai and you don't want to clutter your Vault so this requires a bit of deliberate action on your side to save what you find to save it locally and finally let's look at what it takes to get started with this so first of all you need to install xol AI from the script Library how does it work you click here to open the obsidian tools panel and top left you see install or update excal draw scripts if I open this then I can search for exol AI like this it will take me here and if I click on this then here I can just simply click install now it's already installed so I didn't I don't need to install it but this is how I would install it if I click this it will say installed xol AI once you have exol ai working you need to go to open Ai and you need to create an account and add money to that account so this is the link you need to open and you can see here that this is my balance I have $392 by the way generating images we can look at how much today's usage cost so generating all the images we did right now I was already I think at 9 cents so today this image generation for the video cost me about 40 cents I don't think it's a huge amount of money so this is there you can see the usage information as well once you have your account and you've added money to the account you need to add at minimum $5 otherwise the GPT Vision model is not going to be available to you if you have at least $5 or on your account you still need to wait about 10 20 minutes for open AI to activate your access and then you can come here and configure in obsidian you need to create an open AI API key you can see the link right here if I click on this it takes me to the API keys I can click here to create a new secret key and then the key is going to be created and then I need to go to plug-in settings right here and I need to click on EXC colar draw AI settings and here I can configure my API key and my different models as well and finally when you're done come here to the tools panel and find exol AI here amongst the downloaded scripts if you long click on it then you can add it pin it to the side panel and from then on it's going to work as I've demonstrated it to you uh in the video so that's all I wanted to share with you today I already used it a lot in my workflows I enjoy it so much yes it costs a little money but the amount is not so much and I think the benefit you're getting with the images with the mind maps with the workflows is super valuable give it a try and let me know what you think in the comments below thank you, ExcaliAI Enhanced: More Visual Thinking Power, Zsolt's Visual Personal Knowledge Management, 01hjbnw2an6bbyr9s9gpxdv992.