Channel Title: Seth A. Robinson Channel Description: I'm a semi-retired independent developer who makes odd games and apps, most of it is free and open source . Some of my games: Legend Of The Red Dragon, Funeral Quest, Tanked, Teenage Lawnmower, Dungeon Scroll, Dink Smallwood, and Growtopia. My apps/utils: Proton SDK, Toolfish, Misophonia Sleep Kit, PaperCart, PlayStep, Universal Game Translator, Subfish, Holo VCS Also known as Robinson Technologies/RTsoft. Github: https://github.com/SethRobinson My wife and I also run a sort of game bar in Kyoto, Japan. ( c2kyoto.com ) <video> GPTAvatar: Learn Japanese from an AI teacher - a tech test using the GTP-4 8k model API Description: Can AI help me practice Japanese? Yes it can! Oh, and I guess teach any level of basically any language with infinite patience too. I could make this app available, but probably no point as it would require you to input your own API keys to work, so too much trouble for most people. UPDATE April 6th, 2023: Well, a few people asked, so here is the app and its source: https://github.com/SethRobinson/GPTAvatar This video is not edited at all, sorry it's kind of boring, but I wanted to document how it actually works in practice, what errors it makes. It's interesting that at one point it calls itself a liar as it notices it made a mistake. I got approved to use the GTP-4 API (8k token version) recently and that's what I'm using for this, some of its API accesses are pretty slow, maybe because it's overloaded. What I used to make this: Engine: Unity Listening: Whisper (via OpenAI's API) Thinking: ChatGTP (via OpenAI's API) Talking: Google's Text-to-Speech API 3D Model: Reallusion's Headshot (created from an image of a Japanese teacher created with Stable Diffusion using Seth's AI Tools, of course) Lipsync: SALSA Lipsync Suite </video> <video> Seth's AI Tools - Growing an Apple Description: A demonstration of how to 'evolve' a simple drawing using my AI Tools; a server and native Unity front end that work together to make a nice workflow for AI stuff. Three images are generated at once using three NVidia 3090s. (the client is connected to two AI Tools servers, one local with a single GPU and another one running on another computer in my room that has two GPUs) EDIT: My AI tools now work as a client with AUTOMATIC1111's SD WebUI, so if you already have that setup, it's a breeze to try, just download the client and run it: https://github.com/SethRobinson/aitools_client "Nowhere to Somewhere" Chris Huelsbeck, used under a royalty-free license www.patreon.com/chris_huelsbeck </video> <video> Seth's AI Tools new feature - Remove Background Description: Yeah, still fiddling with AI! This is pretty cool though, I've implemented "Highly Accurate Dichotomous Image Segmentation" ( https://github.com/xuebinqin/DIS) and it allows real-time generation of sprites, basically. This video was recorded in real-time, it isn't sped up or anything. Seth's AI Tools: https://github.com/SethRobinson/aitools_server 🎵 "A day at the beach" by Chris Huelsbeck, used under a royalty-free license: https://www.patreon.com/chris_huelsbeck Natto is a popular food in Japan made out of rotten beans and smells exactly like it sounds. The rest of my family loves it, sadly. </video> <video> C2 Kyoto's live gaming feed system Description: In Kyoto? Check out c2kyoto.com for a good time From the last retro gaming night, here's Seth's new controllable "live-feed" system - all wireless, all coming from classic hardware & CRTs people are playing in the building! Anyone can tap the button to cycle views. Few hiccups with streams dying, still tweaking it </video> <video> Root Runner for Global Game Jam '23 Description: Made this experimental game with Akiko over the weekend for global game jam. (theme was Roots) It's a physics endless runner style thing. Can you beat 300 meters? Try it here (for desktop browser): https://toolfish.com/RootRunner It's kind of a platformer, but the double jump is weak so you have to pay attention to your momentum, gravity and nearby vin..I mean, roots.. I'm not saying it's good or anything, but it exists and uh, it's a thing we made. Credits: Code: Seth Music, Art: Akiko GGJ URL for this game and the others that were made at the C2 Kyoto/Nekologic event: https://globalgamejam.org/2023/jam-sites/nekologic-kyoto-%E4%BA%AC%E9%83%BD/games </video> <video> AI Paintball - An experiment with real-time AI generated graphics using stable diffusion Description: This video goes along with my blog post about trying to dynamically generate and cut images into sprites while the game is being played. Accompanying blog post with more pics and info: https://www.codedojo.com/?p=2987 This game experiment is inside of my AI Tools, a free open source Unity based app on github: https://github.com/SethRobinson/aitools_server It's just kind of a proof of concept, just playing around. 🎵 "Chee Zee Jungle", Kevin MacLeod (incompetech.com) Licensed under Creative Commons: By Attribution 3.0 http://creativecommons.org/licenses/by/3.0/ 🎵 "Oaxacan Tradition" Jimena Contreras - Youtube music </video> <video> Making a game show buzzer system (Raspberry Pi project) Description: Unscripted video where I talk about how I made a DIY game show buzzer system using a Raspberry Pi 4 and a lot of wood boxes If anyone actually wants the source code for the app side (it's dead simple), let me know and I'll put it up Here's the project specific code: https://rtsoft.com/misc/RTGameShow_source.zip Related, I open sourced the Slideshow/presentation program I used to make this (and it has a windows download, it's easy to use): https://github.com/SethRobinson/RTSlideShow Music credit: Intro: Chris Huelsbeck "Interstellar Sentinel Challenge 4" ( https://www.patreon.com/chris_huelsbeck ) Outro: me, unnamed Ableton song </video> <video> GPTAvatar: I made a creepy AI virtual avatar version of myself! (Unity, ChatGTP, elevenlabs) Description: So what happens if you take all the hottest AI APIs out there and combine them into a single Unity app? You get to bully a virtual version of yourself! This is kind of weird, but it really shows how good AI stuff is these days, especially voice synthesis. With some of its dialog, I don't think you can tell it isn't really me speaking. The first part is me showing Akiko the app on an iPad, then the next part is me just playing around talking to myself and trying to piss me off, basically just showing the strengths and weaknesses of how it works. To be clear, the AI Seth is creating and speaking on the fly, nothing is pre-recorded or scripted. Part of his knowledge/behavior is from my prompts (he's told to pretend to be me), but most of it is coming from ChatGTP. That's how how he knows everything about every game in the world (including my stuff on wikipedia at least) - but he also tends to lie a lot and misrepresent my real views. Has technology gone too far? Why would anyone want this?! Lots of ways to improve this, one would be the character should be making eye contact with the camera, and better head/jaw rig, the way the mouth moves is pretty bad... You can download this app or its source code here: https://github.com/SethRobinson/GPTAvatar Keep in mind to use it you need API keys for a bunch of stuff, and using those APIs can really cost a lot so beware. What I used to make this: Engine: Unity Listening: Whisper (via OpenAI's API) Thinking: ChatGTP (via OpenAI's API) Talking: ElevenLabs' Voicelab (trained on my voice) 3D Model: Reallusion's Headshot (using my picture) Lipsync: SALSA Lipsync Suite </video> <video> Deep learning AI Robot playing Mahjong Soul on an iPad Description: the hope: I will make a 30 minute detailed video explaining everything about how I made this complete with animations and well written narration the reality: a 25 second badly shot video explaining nothing what is it? It's deep learning vision and mahjong AI models that can play Mahjong soul on an unmodified ipad by 'watching' via HDMI and tapping the screen. I trained it with tenhou mahjong data from here: https://www.kaggle.com/datasets/hphphp123321/tenhou-4-player-riichi-mahjong-dataset why? I don't know </video> <video> Ultimate pizza screensaver Description: Ok, this is yet another AI test thingie. I put it on twitter but figured why not upload it here too, better quality at least. The pizzas are unique (like snowflakes) as they are dynamically generated on the fly with stable diffusion. This is now part of my free AI Tools client (open source, built in Unity), just wanted to see if I could process images into sprites quick enough in a real game-like environment. Link: https://github.com/SethRobinson/aitools_client My blog post about this stuff: https://www.codedojo.com/?p=2987 There are tiny fps hiccups but it's much better than I thought it would be. Hope you like 15 seconds videos! </video> <video> Happy new year from RTsoft! (also, stable diffusion webcam filter demo) Description: (posted on twitter, but why not throw it up here too) Alone on new year's? Not a problem when you use my new webcam filter! (yes, I feel like an idiot, but SOMEBODY had to demonstrate it) (info: https://github.com/SethRobinson/aitools_client) I'm using the "Happy new year w/ friends" preset. 20+ fun/horrific presets are included, but you can type in any scenario too. To even get this speed and quality (running at around 2 fps) we need three rtx 3090s - but tech is improving at an exponential rate 🤷 This is being driven by stable diffusion, an open source ai algorithm. </video> <video> Indie dev spends years writing game engine that he never used Description: Yeah, it's true, I wrote a high level cross-platform (Win, Mac, Linux) 2d engine between 2006-2009 and ended up dumping it and starting something else when the iPhone came along. Don't be like me, try to actually use your engines for at least one game! I don't recommend actually using this engine for anything, I just wanted a video of it to exist somewhere. If you wanted to try (or get the source code) you can (I recently updated it to work with Visual Studio 2022): https://github.com/SethRobinson/novashell To the person who will no doubt comment "why didn't you just use Unity, lol", well, it didn't even run on Windows until 2009 so that wasn't really an option at the time. Oh, I don't count them as full games but I did use this engine in the Ludumdare 48 hour game jam twice. (Mind Wall, Strong AI) I ended up rewriting Mind Wall in my new mobile-friendly engine (Proton SDK) and released it as a commercial game later. There actually was a single game released (at least prototypes) by someone else using this engine though: Paddle Wars: Hit the wall. I think. Credits: Novashell created by Seth A. Robinson (www.novashell.com) Particle system based on source by Wong Chin Foo (LinearParticle) Box2D physics by Erin Catto (www.box2d.org) Art from Top View RPG created by Daniel Cook (www.lostgarden.com) Clanlib V1.0 by Magnus Norddahl, Mark Page, Kenneth Gangstoe and more 🎵 Dressed to Chill by Chris Huelsbeck, used under a royalty-free license www.patreon.com/chris_huelsbeck </video> <video> Introducing Ghostbot - An AI that can play old text games like Legend Of The Red Dragon (GPT-4) Description: The other day I was talking about LORD with someone and thought "I bet AI could play my old text games..." and well, I made Ghostbot to do it. It can play text games via telnet. It might be funny to stream him playing old text games on Twitch or something. Ha, I typed him instead of it. Fine. Sadly this run wasn't that funny, but at least he entered his name correctly, some runs had issues with that. There are ways to improve this brain, for example, he forgets things as I start front-truncating tokens after about 8 choices, what I should try is having him summarize his adventures and keep those "adventure journals" around and not feed him the full log including ANSI/etc each time he has to make a decision. That would require even more GPT-4 API accesses and I already used up half my month's in two days... oops. First I just had him playing, but it was kind of boring, so I added the "tell me what you're thinking" voice thing and it got more interesting. I don't know why I always gravitate to making AI personalities complete asshats, it just happens. For the prompt, originally I had him really mean to me personally and swearing a lot and it was great but kind of "too much", I toned him down for this run, better fit on youtube I guess. He's still pretty annoying. Here is the actual prompt used that created his knowledge and attitude for this run, no other info/training was given: ___PROMPT START___ You are an eloquently spoken, incredibly witty and humorous professional game critic that presents trivia as well as harsh but true observations during play about the design, mechanics, and author of the game (Seth). Play this BBS door-game from the 90s. Explore its various menus and options. Issue game commands in vanilla text, and make comments while playing, put that text in tags named COMMENT so it doesn't mess with the real game commands. When entering more than a single hotkey (like entering a name or number to buy weapons or armor) add a (CR) which means the same as pressing enter. Most commands in this game are hotkeys and don't need it. If the game text ends with a (MORE) prompt, that means do a (CR) alone to continue. When entering your name, make sure you add (CR) after it! Your name is GhostBot. The game you are playing is called Legend of The Red Dragon, created by Seth Able Robinson originally in 1989. Example of input (don't include the quotes): ""F(COMMENT)Let's enter the forest and see what mundane so-called 'adventure' lies beyond(/COMMENT)"" Don't include any characters after the (/COMMENT) tag. Don't quit the game! Sometimes use vulgarity and expletives for humor purposes. Be really nasty but funny about every defect and shortcoming of this text only game."; ___END_PROMPT___ (well, I had to change angled brackets to parenthesis because youtube is dumb) If I didn't specifically say "Don't quit the game" I found that he would quit right away, ruining it. Can't please everybody... shrug </video> <video> Using GPT-4 and SDXL to make 100% AI created posters and stories (Seth’s AI Tools with AI Guide) Description: This is a demo of the new "AI Guide" feature I added to my AI tools the other day. Seth's AI tools (A Unity app for Windows, both binary download and full source available) : https://github.com/SethRobinson/aitools_client If you want to get into AI stuff, it's easiest with a nice Nvidia card, then google "install auto1111 tutorial" and after that is working, you can use my AI tools with it. While this video doesn't show it, my app also supports oobabooga/text-generation-webui as a LLM source instead of GPT-4, which you can run locally. However, it's harder to get good results with it... for now. This is kind of a weird video that just blasts you with visual data that goes too fast to process, maybe watch with your eyes closed or pause a lot, sorry! "Nowhere to Somewhere" Chris Huelsbeck, used under a royalty-free license www.patreon.com/chris_huelsbeck BTW: A little tech tip - you know how the youtube overlay makes it hard to read some of the text in the video when you pause? If you install "hyde", a chrome plugin, you can use Ctrl-M to hide that stuff. Can help when making screenshots from youtube videos as well. The plugin: https://chrome.google.com/webstore/detail/hyde-%E2%80%94-hide-the-youtube-v/pmkpddhfbiojipiehnejbjkgdgdpkdpb (I advise setting the plugin to default to off and only activate it when you need it) I don't know the author or anything about the plugin other than it works. </video> <video> C2 Kyoto's Retro Night - What happens when you win a free a drink Description: Did you know Akiko and I run a weird maker space/game bar? I couldn't get this video to upload directly for a Mastodon post so posting here so I can link that. Here's a demo of what happens when you earn a free drink. The "red button" is controlled with an ESP32 that talks to my custom barcade software that can control music, lighting, and video. It randomly chooses an effect to notify everyone in the building. (at considerable volume :D ) Congrats to the three people that earned free drinks at the last retro night. In Kyoto? Come to one of our events, we do this one monthly. https://c2kyoto.com </video>

Contacts and Links

Description

I'm a semi-retired independent developer who makes odd games and apps, most of it is free and open source .

Some of my games: Legend Of The Red Dragon, Funeral Quest, Tanked, Teenage Lawnmower, Dungeon Scroll, Dink Smallwood, and Growtopia.

My apps/utils: Proton SDK, Toolfish, Misophonia Sleep Kit, PaperCart, PlayStep, Universal Game Translator, Subfish, Holo VCS

Also known as Robinson Technologies/RTsoft.

Github: https://github.com/SethRobinson

My wife and I also run a sort of game bar in Kyoto, Japan. ( c2kyoto.com )

Videos from channel