Step-by-step AI workflows with ComfyUI, Blender and open-source models for images, 3D, video and voice, with the templates to run them.https://tenitsky.com/https://tenitsky.com/guides/ai-room-3d-prop-showroom-blender/https://tenitsky.com/guides/ai-room-3d-prop-showroom-blender/Generate a photoreal room from a Blender blockout with Stable Diffusion and a depth ControlNet, project it back onto the geometry, then place a real 3D model inside it with matched lighting, a shadow catcher, light linking, bloom and an animated depth-of-field camera move.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/artist-style-transfer-3d-scenes-blender/https://tenitsky.com/guides/artist-style-transfer-3d-scenes-blender/Pose a free Mixamo character in Blender, block out a scene with cubes, and turn the render into a finished image in an artist’s style using Flux, a style LoRA and a depth ControlNet.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/auto-crop-photos-comfyui/https://tenitsky.com/guides/auto-crop-photos-comfyui/An intelligent batch crop: Florence 2 finds the face (or anything you name) in every photo of a folder, and ComfyUI crops a clean, centred frame around it, in any aspect ratio.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/automated-3d-to-concept-art-pipeline/https://tenitsky.com/guides/automated-3d-to-concept-art-pipeline/Turn a folder of 3D blockout screenshots into concept art with Flux: depth ControlNet, prompt lists, prompt stacking, a prompt per object, and automatic portrait/landscape sorting, built up from simple workflows to a full overnight pipeline.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/batch-concept-art-from-photos-flux/https://tenitsky.com/guides/batch-concept-art-from-photos-flux/Turn a whole folder of reference photos into consistent AI concept art: Flux 1 dev, a Depth Anything V2 depth map and Flux depth ControlNet v3, with file names that link every result back to its photo.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/blockout-to-render-blender-stable-diffusion/https://tenitsky.com/guides/blockout-to-render-blender-stable-diffusion/Block out a lighthouse scene from cubes in Blender 4.3, render a depth pass, turn it into a photoreal image with Stable Diffusion and a depth ControlNet, then project the render back onto the geometry for a short camera move with animated water.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/comfyui-manager-image-to-image/https://tenitsky.com/guides/comfyui-manager-image-to-image/Install ComfyUI-Manager with git so you can add custom nodes and fix missing ones, then build an image-to-image workflow that merges a skull and a teapot by tuning one value: denoise.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/comfyui-node-basics-text-to-image/https://tenitsky.com/guides/comfyui-node-basics-text-to-image/Install ComfyUI portable on Windows, download a Stable Diffusion 1.5 checkpoint, and build the basic text-to-image graph from an empty canvas: Empty Latent Image, Load Checkpoint, prompts, KSampler, VAE Decode and Save Image, with what each setting does.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/concept-car-design-blender/https://tenitsky.com/guides/concept-car-design-blender/My full process for a futuristic one-seater, in three parts: a fast low-poly concept built with modifiers, booleans and materials, the design decisions behind it, and the retopology into a clean subdivision model with a custom logo.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/control-ai-pc-from-browser/https://tenitsky.com/guides/control-ai-pc-from-browser/Use your powerful AI PC from a laptop in another country without remote-desktop lag: Tailscale for a private network, one change to ComfyUI’s start file, and MEGA to sync your renders.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/fix-ai-faces-hands-automatically/https://tenitsky.com/guides/fix-ai-faces-hands-automatically/A batch workflow that finds the hands and faces in a folder of AI images, re-renders just those areas at higher resolution with Flux, and saves the fixed images: Florence 2 masking plus the Impact Pack’s SEGS detailer.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/fix-ai-hands-comfyui/https://tenitsky.com/guides/fix-ai-hands-comfyui/Mask a broken hand, let Flux redraw it at higher resolution with the SEGS detailer, and use a quick paint-over to fix the hardest cases. Includes the denoise values that work.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/florence-2-masking-comfyui/https://tenitsky.com/guides/florence-2-masking-comfyui/Type “crown”, “tiger” or “glasses” and Florence 2 returns a mask. Build the workflow in ComfyUI, inpaint the masked area with SD 1.5, combine two masks into one, and turn a mask into a border.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/free-ai-upscaling-sd15-comfyui/https://tenitsky.com/guides/free-ai-upscaling-sd15-comfyui/Upscale photos and renders 4× (or 16× by chaining) locally and for free: Ultimate SD Upscale with an SD 1.5 checkpoint and an upscale model from OpenModelDB, tiled at 512 px so it runs on a 4 GB GPU.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/generate-prompts-from-images-comfyui/https://tenitsky.com/guides/generate-prompts-from-images-comfyui/Run a vision language model locally with Ollama, let it describe each sketch and write a prompt, then generate new images with SDXL or Flux and a depth ControlNet. Thousands of variations overnight, all offline.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/inpainting-flux-comfyui/https://tenitsky.com/guides/inpainting-flux-comfyui/Paint a soft mask, let Flux Dev re-render just that area at higher resolution with the SEGS detailer, then save, re-upload and move to the next part. The settings that work, and why small masks beat big ones.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/introduction-to-runpod/https://tenitsky.com/guides/introduction-to-runpod/Why I moved ComfyUI from my own RTX A5000 to RunPod, what cloud GPUs cost, which GPU to pick for each job, and the setup from a network volume to your first render.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/ltx-2-5-talking-avatar/https://tenitsky.com/guides/ltx-2-5-talking-avatar/Turn one portrait and a voiceover of any length into a lip-synced talking avatar that keeps the same face, whether the video runs 30 seconds or 50 minutes. Setup, settings and how the chunking works.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/minimax-h3-batch-avatar-videos/https://tenitsky.com/guides/minimax-h3-batch-avatar-videos/One portrait and a folder of voiceovers in, a finished talking-head video for every file out. How the MiniMax H3 batch workflow works, which GPU to use, and what it costs per finished minute.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/nano-banana-concept-art-test/https://tenitsky.com/guides/nano-banana-concept-art-test/I ran my own portfolio (characters, vehicles, environments and old black-and-white concepts) through Google’s Nano Banana image editor. Where it is genuinely useful, where it misfires, and how I prompted it.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/photoreal-pepe-flux/https://tenitsky.com/guides/photoreal-pepe-flux/Turn a flat cartoon into a photoreal character with Flux in ComfyUI: a depth map from the drawing, a ChatGPT-written prompt edited by hand, and a short animation with Seedance on fal.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/prompt-cycler-comfyui-node/https://tenitsky.com/guides/prompt-cycler-comfyui-node/My free custom node turns every line of a text box into its own prompt and moves to the next one on each run. How to install it, connect it, show the current prompt, and add a fixed subject to every line with a Concatenate node.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/prompt-lists-3d-render-art-styles/https://tenitsky.com/guides/prompt-lists-3d-render-art-styles/Load hundreds of prompts from a text file with the Inspire Pack, add a fixed subject and background with Text Concatenate, and pair a folder of 3D screenshots with matching prompts so each image gets the right one.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/qwen3-tts-voiceovers/https://tenitsky.com/guides/qwen3-tts-voiceovers/My RunPod template installs Alibaba’s open-source Qwen3-TTS with four ready workflows. What each one does, how scripts of any length are handled, and what hardware you need.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/real-time-concept-art-blender-stable-diffusion/https://tenitsky.com/guides/real-time-concept-art-blender-stable-diffusion/Pose a Mixamo character in a box-built Blender scene and get a new concept image every time you move the camera. A screen-share ComfyUI setup, from fast SD 1.5 to sharper SDXL Lightning.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/relight-photos-ic-light-comfyui/https://tenitsky.com/guides/relight-photos-ic-light-comfyui/Change the lighting of a portrait, a group photo or a landscape with IC-Light v1 and SD 1.5 in ComfyUI. Build the workflow, keep the face with Detail Transfer, steer the light with masks, and tame burned-out results.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/sci-fi-guns-blender-ai/https://tenitsky.com/guides/sci-fi-guns-blender-ai/Model a prop from boxes in Blender and watch ComfyUI turn it into concept art live, as you move and scale it. Built from scratch: Mixlab screen share, Juggernaut XL Lightning and a depth ControlNet, even on 4 GB of VRAM.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/simple-3d-shapes-to-tank-concepts/https://tenitsky.com/guides/simple-3d-shapes-to-tank-concepts/Turn a rough 3D blockout of a tank into concept art with Flux, then into a textured 3D model with Hunyuan3D and TRELLIS on Hugging Face, and back into Blender. Plus how ChatGPT and Gemini compare for the concept step.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/sketch-to-art-4gb-gpu/https://tenitsky.com/guides/sketch-to-art-4gb-gpu/Build a free, offline sketch-to-image workflow step by step: SDXL with Juggernaut XL, a Depth Anything depth map and an SDXL depth ControlNet. It runs on a laptop with only 4 GB of VRAM.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/sketch-to-art-flux-cloud-gpu/https://tenitsky.com/guides/sketch-to-art-flux-cloud-gpu/Turn a quick pen, Photoshop or Alchemy sketch into polished concept art with Flux and a depth ControlNet. Every node explained, the resolution rule that avoids glitches, and how to run it without a big GPU.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/sort-images-by-aspect-ratio-comfyui/https://tenitsky.com/guides/sort-images-by-aspect-ratio-comfyui/Detect whether each image in a folder is landscape or portrait and size the workflow to match, with no AI model involved. Two versions: an image switch that is easy to follow, and a faster number switch.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/stable-audio-3-sound-effects/https://tenitsky.com/guides/stable-audio-3-sound-effects/A ComfyUI setup that turns a list of prompts, lengths and file names into hundreds of AI sound effects, using Stable Audio 3 Medium. Runs on almost any GPU, locally or on RunPod.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/stock-photos-to-cinematic-art-sdxl/https://tenitsky.com/guides/stock-photos-to-cinematic-art-sdxl/A free, fully local ComfyUI workflow that restyles a whole folder of photos with SDXL and a depth ControlNet. It runs on a graphics card with just 4 GB of VRAM.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/stop-colour-bleeding-comfyui/https://tenitsky.com/guides/stop-colour-bleeding-comfyui/Ask for a white vase with red flowers and the red leaks into the vase and the background. Here is why, and how to keep each part of the prompt separate with weights, Conditioning (Concat) and ConditioningAverage.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/style-transfer-ipadapter-comfyui/https://tenitsky.com/guides/style-transfer-ipadapter-comfyui/Use IPAdapter Plus to carry the look of a reference image into a new generation with SD 1.5: the basic node setup, the weight and weight-type settings that matter, and combining two or three images with Combine Embeds.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/stylized-game-assets-z-image-meshy/https://tenitsky.com/guides/stylized-game-assets-z-image-meshy/Train a style LoRA on 5–10 of your own images, generate hundreds of characters and props in that style with Z-Image Turbo in ComfyUI, and turn the best ones into 3D models with Meshy.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/train-lora-artist-style/https://tenitsky.com/guides/train-lora-artist-style/Three style LoRAs (Renoir, Hokusai and Albrecht Dürer), each from 6 to 15 images and about $2 of training, plus a ComfyUI workflow that runs hundreds of prompts through them overnight.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/trellis-2-batch-3d-models/https://tenitsky.com/guides/trellis-2-batch-3d-models/Turn a folder of images into hundreds of textured 3D models with Microsoft’s open-source TRELLIS 2 in ComfyUI. How the pipeline works, which settings matter, and how to control polygon count and texture size.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/z-image-thousands-of-images/https://tenitsky.com/guides/z-image-thousands-of-images/A private, uncensored ComfyUI setup that generates images in bulk from a list of prompts and saves them straight to your computer, for a fraction of a cent per image on a rented GPU.Tue, 13 Oct 2026 00:00:00 GMThttps://tenitsky.com/guides/z-image-turbo-runpod/https://tenitsky.com/guides/z-image-turbo-runpod/A short setup guide: Z-Image Turbo running in ComfyUI on a rented GPU in about four minutes, with a network volume so you only install it once, and what to do when your GPU isn’t available.Tue, 13 Oct 2026 00:00:00 GMT