<?xml version="1.0" encoding="UTF-8"?>
<rss version="2.0" xmlns:atom="http://www.w3.org/2005/Atom" xmlns:dc="http://purl.org/dc/elements/1.1/">
  <channel>
    <title>DEV Community: simran</title>
    <description>The latest articles on DEV Community by simran (@simran903).</description>
    <link>https://dev.to/simran903</link>
    <image>
      <url>https://media2.dev.to/dynamic/image/width=90,height=90,fit=cover,gravity=auto,format=auto/https:%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Fuser%2Fprofile_image%2F4211462%2Ff4d6e286-440a-405c-b030-72517ec4b9a7.png</url>
      <title>DEV Community: simran</title>
      <link>https://dev.to/simran903</link>
    </image>
    <atom:link rel="self" type="application/rss+xml" href="https://dev.to/feed/simran903"/>
    <language>en</language>
    <item>
      <title>Can Krea 2 Turbo Create Product Ads? I Tried 5 Text-to-Image Experiments</title>
      <dc:creator>simran</dc:creator>
      <pubDate>Thu, 29 Oct 2026 15:46:35 +0000</pubDate>
      <link>https://dev.to/hexgrid_cloud/can-krea-2-turbo-create-product-ads-i-tried-5-text-to-image-experiments-38il</link>
      <guid>https://dev.to/hexgrid_cloud/can-krea-2-turbo-create-product-ads-i-tried-5-text-to-image-experiments-38il</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;Krea 2 Turbo offers an interesting way to explore these aspects through text-to-image generation. Using ComfyUI, five product-advertising examples demonstrate how the model interprets different prompts and translates creative directions into advertising visuals.&lt;/p&gt;

&lt;p&gt;This article covers the prompts used for each product, the generated results, and the creative decisions behind them. Readers will get a better understanding of how to structure prompts for different advertising styles, guide product placement and lighting, and use Krea 2 Turbo to create visuals suited to different commercial applications.&lt;/p&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fogvebosmf0i1i9zaotm3.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fogvebosmf0i1i9zaotm3.png" alt=" " width="800" height="332"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  Model Configuration and Inference Optimizations
&lt;/h2&gt;

&lt;p&gt;For our image generation experiments, we used &lt;strong&gt;Krea 2 Turbo&lt;/strong&gt; in ComfyUI, combining an INT8-quantized diffusion model with an FP8 text encoder to reduce memory requirements and make inference more efficient.&lt;/p&gt;

&lt;p&gt;Our generation setup consisted of:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;&lt;strong&gt;Component&lt;/strong&gt;&lt;/th&gt;
&lt;th&gt;&lt;strong&gt;Configuration&lt;/strong&gt;&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Diffusion Model&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Krea 2 Turbo — INT8 ConvRot&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Text Encoder&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Qwen3-VL 4B — FP8 Scaled&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;VAE&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Qwen Image VAE&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Inference Framework&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;ComfyUI&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h3&gt;
  
  
  Why These Optimizations Matter
&lt;/h3&gt;

&lt;p&gt;The &lt;strong&gt;INT8 ConvRot diffusion model&lt;/strong&gt; reduces model-weight memory requirements compared with higher-precision variants, making Krea 2 Turbo more practical to run on GPUs with limited VRAM.&lt;/p&gt;

&lt;p&gt;The &lt;strong&gt;Qwen3-VL 4B FP8 Scaled text encoder&lt;/strong&gt; further reduces the memory footprint of prompt processing, while the &lt;strong&gt;Qwen Image VAE&lt;/strong&gt; handles image encoding and decoding.&lt;/p&gt;

&lt;p&gt;Together, these components provide a memory-efficient configuration for exploring Krea 2 Turbo's image generation capabilities in ComfyUI. Actual generation speed, VRAM usage, and output quality depend on the GPU, resolution, and sampling settings.&lt;/p&gt;

&lt;h2&gt;
  
  
  Prompting Approach
&lt;/h2&gt;

&lt;p&gt;For product ads, describe the object's physical construction and finish, then decide where it sits in the frame. A single hero product is usually easier to art-direct than a crowded scene. Keep the headline short, put exact copy in quotation marks, and specify a clear text area. If exact packaging, logos, prices, or legal copy matter, add them in a design editor after generation.&lt;/p&gt;




&lt;h2&gt;
  
  
  1. Premium Headphones - Minimalist Luxury
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Format:&lt;/strong&gt; 4:5 portrait&lt;br&gt;
&lt;strong&gt;Creative direction:&lt;/strong&gt; Warm ivory, matte black, brushed metal, soft studio light.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Prompt&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create a premium commercial advertisement for fictional VANTA wireless over-ear headphones. Show one pair of matte-black headphones with brushed-aluminum hinges and soft memory-foam ear cushions, floating at a subtle three-quarter angle above a warm ivory podium. Place the product in the right two-thirds with a controlled soft shadow. Use large-format studio photography, an 85mm lens, soft directional light from the upper left, realistic metal reflections, and a warm-gray and charcoal palette. Leave the left third clean for typography. Add the exact bold black headline "TUNE OUT THE NOISE", the smaller line "Sound, on your terms." and a minimal CTA reading "EXPLORE VANTA". Keep the spacing sophisticated and the product geometry precise. No other writing or decorative objects.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;What to evaluate:&lt;/strong&gt; Headphone geometry, reflections, silhouette clarity, and whether the left-side copy area remains uncluttered.&lt;/p&gt;

&lt;h3&gt;
  
  
  Output Image
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fi8us7a0jv5e94sa1u4mo.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fi8us7a0jv5e94sa1u4mo.png" alt=" " width="800" height="1065"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  2. Reusable Water Bottle - Performance Advertising
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Format:&lt;/strong&gt; 1:1 square&lt;br&gt;
&lt;strong&gt;Creative direction:&lt;/strong&gt; Electric cobalt, orange contrast, frozen water motion.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Prompt&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create a high-energy advertisement for fictional reusable-bottle brand ROVE. A large brushed stainless-steel bottle with an electric-cobalt finish and a simple screw cap is suspended diagonally in mid-air, surrounded by crisp water splashes and translucent droplets. Use a saturated orange studio background, high-speed flash photography, believable reflections, and dynamic but controlled motion. Keep the bottle as the unmistakable hero. At the top, add the exact white headline "TAKE IT EVERYWHERE", followed by "Meet your everyday bottle." At the bottom, place a clean white CTA reading "SHOP ROVE". Use modern athletic-brand art direction, strong color contrast, and an uncluttered composition. No extra products or text.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;What to evaluate:&lt;/strong&gt; Whether the bottle remains legible against the energetic background and whether water effects support rather than obscure the product.&lt;/p&gt;

&lt;h3&gt;
  
  
  Output Image
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjl4jw0svzu0mxuilep9k.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fjl4jw0svzu0mxuilep9k.png" alt=" " width="800" height="800"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  3. Designer Desk Lamp - Premium Homeware
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Format:&lt;/strong&gt; 16:9 landscape&lt;br&gt;
&lt;strong&gt;Creative direction:&lt;/strong&gt; Warm lamp glow against cool twilight.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Prompt&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create an elegant wide-format advertisement for fictional lighting brand FORM. A sculptural mushroom-shaped brushed-aluminum table lamp sits on a dark walnut desk in a contemporary minimalist apartment at twilight. The lamp casts a warm pool of light over a small open notebook while cool blue evening light enters through a large window. Use architectural lifestyle photography, realistic material detail, and restrained luxury styling. Keep the lamp and interior on the right half; leave the left half as a clean, shadowed wall. Add the exact off-white serif headline "LIGHT CHANGES EVERYTHING", the small uppercase line "OBJECTS FOR LIVING", and the CTA "DISCOVER FORM". No other text.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;What to evaluate:&lt;/strong&gt; The balance between warm and cool light, believable illumination, and negative space for the headline.&lt;/p&gt;

&lt;h3&gt;
  
  
  Output Image
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ff47hpien7kyz0sjf0ps9.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ff47hpien7kyz0sjf0ps9.png" alt=" " width="799" height="451"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  4. Travel Luggage - Lifestyle Conversion Ad
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Format:&lt;/strong&gt; 4:5 portrait&lt;br&gt;
&lt;strong&gt;Creative direction:&lt;/strong&gt; Sage-green luggage, sunlit modern station, aspirational travel.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Prompt&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create a premium travel-lifestyle advertisement for fictional AERA carry-on luggage. Show a muted sage-green hard-shell suitcase with vertical grooves and polished black wheels standing beside a traveler wearing cream tailored trousers and white sneakers; show only the traveler's lower body. Set the scene in a sunlit modern European train station with pale stone architecture and soft distant motion blur. Use natural morning light, editorial 35mm photography, subtle film grain, and quiet-luxury color grading. Keep the suitcase fully visible in the lower center and leave a softly blurred cream area above it. Add the headline "GO SOMEWHERE NEW", the smaller line "Travel looks good on you." and the CTA "MEET AERA". No unrelated typography.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;What to evaluate:&lt;/strong&gt; Suitcase proportions, wheel and handle consistency, and whether the background supports the product without distracting from it.&lt;/p&gt;

&lt;h3&gt;
  
  
  Output Image
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9kx2e7u1bqb0earh5jh4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F9kx2e7u1bqb0earh5jh4.png" alt=" " width="800" height="1065"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  5. Portable Power Bank - Bold Technology Launch
&lt;/h2&gt;

&lt;p&gt;&lt;strong&gt;Format:&lt;/strong&gt; 4:5 portrait&lt;br&gt;
&lt;strong&gt;Creative direction:&lt;/strong&gt; Obsidian black, electric cyan, precise 3D materials.&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;Prompt&lt;/strong&gt;&lt;/p&gt;

&lt;blockquote&gt;
&lt;p&gt;Create a futuristic product advertisement for fictional PULSE portable power bank. Show one precisely shaped rectangular black power bank with rounded edges, a subtle cyan indicator light, and a braided charging cable floating above a glossy obsidian platform. Use translucent glass panels with controlled cyan illumination for depth, ray-traced reflections, crisp rim lighting, and a rich black-and-cyan palette. Center the product slightly below the middle. Add the bold geometric headline "POWER IN YOUR POCKET", the supporting line "Built for the everyday." and a cyan CTA reading "DISCOVER PULSE". Keep exactly one main device and avoid random numbers or extra text.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;p&gt;&lt;strong&gt;What to evaluate:&lt;/strong&gt; Device geometry, cable plausibility, reflections, and the visual hierarchy between product and headline.&lt;/p&gt;

&lt;h3&gt;
  
  
  Output Image
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5lev0cqkivtcjnfs8twj.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F5lev0cqkivtcjnfs8twj.png" alt=" " width="800" height="1065"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;Krea 2 Turbo on&lt;/strong&gt; &lt;a href="http://HexGrid.cloud" rel="noopener noreferrer"&gt;&lt;strong&gt;&lt;u&gt;HexGrid.cloud&lt;/u&gt;&lt;/strong&gt;&lt;/a&gt;
&lt;/h2&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;&lt;em&gt;Run&lt;/em&gt; Krea 2 Turbo *on HexGrid in one click — no CUDA setup, no dependency hunting, no environment debugging.&lt;/strong&gt;*&lt;/p&gt;

&lt;p&gt;&lt;em&gt;The template comes preconfigured with ComfyUI, the required runtime, and the&lt;/em&gt; &lt;strong&gt;Krea 2 Turbo&lt;/strong&gt; &lt;em&gt;setup so you can launch a GPU workspace and start generating immediately.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;em&gt;No manual CUDA installation. No PyTorch version matching. No broken requirements. No hours spent wiring the environment together.&lt;/em&gt;&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Just choose the&lt;/em&gt; &lt;strong&gt;Krea 2 Turbo&lt;/strong&gt; &lt;em&gt;template, start the instance, and open ComfyUI.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Pricing starts from&lt;/em&gt; &lt;strong&gt;&lt;em&gt;$0.58/hour&lt;/em&gt;&lt;/strong&gt;*, so you can spin up a GPU only when you need it and shut it down when you’re done.*&lt;/p&gt;
&lt;/blockquote&gt;

</description>
      <category>krea2</category>
      <category>comfyui</category>
      <category>productadvertising</category>
    </item>
    <item>
      <title>Testing FastH3 Omni in ComfyUI: 5 Video Experiments on an NVIDIA L40</title>
      <dc:creator>simran</dc:creator>
      <pubDate>Tue, 27 Oct 2026 18:35:23 +0000</pubDate>
      <link>https://dev.to/hexgrid_cloud/testing-fasth3-omni-in-comfyui-5-video-experiments-on-an-nvidia-l40-4ggp</link>
      <guid>https://dev.to/hexgrid_cloud/testing-fasth3-omni-in-comfyui-5-video-experiments-on-an-nvidia-l40-4ggp</guid>
      <description>&lt;h2&gt;
  
  
  Introduction
&lt;/h2&gt;

&lt;p&gt;Getting an AI video model to generate a good-looking clip is one thing. Getting it to keep the same character, preserve product details, and follow a scene across multiple shots is a different challenge.&lt;/p&gt;

&lt;p&gt;I wanted to see how FastVideo FastH3 Omni Reference handles these situations in practice, so I ran five experiments in ComfyUI on an NVIDIA L40 with 48 GB of VRAM. The tests cover everything from a short fashion film and a luxury product advertisement to character consistency, a sci-fi scene, and a multi-shot commercial.&lt;/p&gt;

&lt;p&gt;Rather than judging the model by a single output, I used different reference images and prompts to see how it handles identity, clothing, objects, camera movement, and scene composition. I've included the prompts, generation settings, and resulting videos so you can try similar experiments yourself.&lt;/p&gt;




&lt;h2&gt;
  
  
  Model Configuration and Inference Optimizations
&lt;/h2&gt;

&lt;p&gt;For our five product advertising experiments, we used the &lt;strong&gt;FastVideo FastH3 OmniRef 8-step model&lt;/strong&gt; in ComfyUI, with quantized model weights and optimized attention to make reference-to-video generation more memory-efficient.&lt;/p&gt;

&lt;p&gt;Our generation setup consisted of:&lt;/p&gt;

&lt;div class="table-wrapper-paragraph"&gt;&lt;table&gt;
&lt;thead&gt;
&lt;tr&gt;
&lt;th&gt;Component&lt;/th&gt;
&lt;th&gt;Configuration&lt;/th&gt;
&lt;/tr&gt;
&lt;/thead&gt;
&lt;tbody&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Diffusion Model&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;FastH3 OmniRef 8-Step Pruned INT8 ConvRot&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Text Encoder&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;Qwen3-VL 32B MiniMax H3 (NVFP4 AWQ)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Video VAE&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;MiniMax H3 INT8 ConvRot&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Audio VAE&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;MiniMax H3 FP32&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Video Sampling Shift&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;12.0&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Audio Sampling Shift&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;3.0&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Attention Backend&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;ComfyKitchen Attention&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;Sparse Attention&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;VSA (Video Sparse Attention)&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;VSA Keep Percentage&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;10%&lt;/td&gt;
&lt;/tr&gt;
&lt;tr&gt;
&lt;td&gt;&lt;strong&gt;VSA Active Range&lt;/strong&gt;&lt;/td&gt;
&lt;td&gt;0.0–1.0 (entire sampling schedule)&lt;/td&gt;
&lt;/tr&gt;
&lt;/tbody&gt;
&lt;/table&gt;&lt;/div&gt;

&lt;h3&gt;
  
  
  Why These Optimizations Matter
&lt;/h3&gt;

&lt;p&gt;The &lt;strong&gt;INT8 ConvRot diffusion model and video VAE&lt;/strong&gt; reduce model-weight memory requirements compared with their higher-precision counterparts, while the &lt;strong&gt;NVFP4 AWQ text encoder&lt;/strong&gt; further reduces the memory footprint of the large Qwen3-VL 32B encoder.&lt;/p&gt;

&lt;p&gt;We also configured &lt;strong&gt;ComfyKitchen Attention alongside VSA with a 10% keep ratio&lt;/strong&gt;, enabling sparse attention throughout the sampling schedule. This is intended to reduce attention computation and improve inference efficiency, although the actual performance gains depend on the hardware and workload.&lt;/p&gt;

&lt;p&gt;Combined with FastH3's &lt;strong&gt;8-step distilled architecture&lt;/strong&gt;, these optimizations provide a practical configuration for experimenting with high-quality, reference-guided product advertisements without relying on the full-precision MiniMax H3 pipeline.&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;Generation 1: Streetwear Fashion Film — 5 seconds&lt;/strong&gt;
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Input Image 1
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fengxrz7oyo3pkecit1g7.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fengxrz7oyo3pkecit1g7.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Input Image 2
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fw1rg97upiq9l6r3q1414.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fw1rg97upiq9l6r3q1414.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Input Image 3
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fch7usq2b30cm6d2wr9b6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fch7usq2b30cm6d2wr9b6.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Prompt
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;subject_definitions: &amp;lt;Subject 1&amp;gt; is the fictional adult fashion model in &amp;lt;Picture 1&amp;gt;. &amp;lt;Picture 2&amp;gt; is the specific oversized black puffer jacket that &amp;lt;Subject 1&amp;gt; wears. &amp;lt;Picture 3&amp;gt; is the brutalist concrete hall used as the location.&lt;/p&gt;

&lt;p&gt;summary: [reference generation] A five-second, single-take high-fashion campaign with &amp;lt;Subject 1&amp;gt; wearing &amp;lt;Picture 2&amp;gt; inside &amp;lt;Picture 3&amp;gt;; powerful, minimal, photorealistic.&lt;/p&gt;

&lt;p&gt;retention_analysis: &amp;lt;Subject 1&amp;gt; (Shot 1): fully_preserved — preserve the face, black choppy hair, facial proportions, relaxed androgynous styling, dark cargo trousers and boots of &amp;lt;Picture 1&amp;gt;. Preserve the puffer jacket's bulky silhouette, horizontal red chest band, quilting, collar and zipper exactly as &amp;lt;Picture 2&amp;gt;. Preserve the bare concrete and tall windows of &amp;lt;Picture 3&amp;gt;. Do not copy the gray studio background from &amp;lt;Picture 1&amp;gt; or &amp;lt;Picture 2&amp;gt;.&lt;/p&gt;

&lt;p&gt;detailed_description: [Shot 1, 00:00–00:05] One continuous live-action full-body fashion shot. &amp;lt;Subject 1&amp;gt; stands at the far end of the hall from &amp;lt;Picture 3&amp;gt;, wearing the jacket from &amp;lt;Picture 2&amp;gt; over the outfit from &amp;lt;Picture 1&amp;gt;. The model takes three slow deliberate steps toward the camera. The camera tracks backward smoothly at hip height, then slowly raises to chest height. The red band stays a solid physical garment detail, not a floating graphic. At the end the model stops and turns their head ten degrees toward the windows. Natural cloth compression and soft shoe contact, realistic restrained movement, cold overcast light, subtle reflections on the concrete floor. No cuts, no extra people, no logos, no text overlays.&lt;/p&gt;

&lt;p&gt;overall_soundscape: A low airy interior room tone, three distinct heavy rubber-sole footsteps, quiet fabric rustling and a distant wind through the windows.&lt;/p&gt;

&lt;p&gt;non_diegetic_music: Sparse industrial sub-bass pulses with spacious pauses; understated, not overpowering the footsteps.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  Setup Configs
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;Red Band — fashion film Duration: 5 seconds; choose 124 frames / 24 fps. Suggested FIRST test output: about 1-megapixels, preserve reference aspect ratio via the official template if possible. Ref_image_size: match. Steps: 8 | guidance: 1.0 | sigma shifts: video 12.0 / audio 3.0 | VSA: 10%&lt;/p&gt;

&lt;p&gt;Connect ordered inputs into ref_images on MiniMaxH3ReferenceToVideo: &amp;lt;Picture 1&amp;gt; -&amp;gt; Picture_1.png -&amp;gt; the protagonist’s face, haircut, base pants and boots &amp;lt;Picture 2&amp;gt; -&amp;gt; Picture_2.png -&amp;gt; the exact black quilted puffer jacket with its broad red horizontal panel &amp;lt;Picture 3&amp;gt; -&amp;gt; Picture_3.png -&amp;gt; the concrete windowed hall and its cold daylight&lt;/p&gt;

&lt;p&gt;Do not connect these images to the First Last Frame to Video node.&lt;/p&gt;

&lt;p&gt;Actual render quality and VRAM requirements will vary.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  Output
&lt;/h3&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/zeKvMvNRG7A" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;Generation 2: Luxury Product Advertisement — 5 seconds&lt;/strong&gt;
&lt;/h2&gt;

&lt;h3&gt;
  
  
  Input Image 1
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpiclxqzc2aly9zdnq8df.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fpiclxqzc2aly9zdnq8df.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Input Image 2
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3kjj3csfouov00otd5uh.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F3kjj3csfouov00otd5uh.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Input Image 3
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frx4qgfw89gypgoruqgkt.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Frx4qgfw89gypgoruqgkt.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  Prompt
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;subject_definitions: &amp;lt;Subject 1&amp;gt; is the fictional adult woman from &amp;lt;Picture 1&amp;gt;. &amp;lt;Object 1&amp;gt; is the exact sculptural chrome wraparound sunglasses from &amp;lt;Picture 2&amp;gt;. &amp;lt;Location 1&amp;gt; is the curved concrete atrium from &amp;lt;Picture 3&amp;gt;.&lt;/p&gt;

&lt;p&gt;summary: [reference generation] A five-second premium eyewear commercial: &amp;lt;Subject 1&amp;gt; models &amp;lt;Object 1&amp;gt; in &amp;lt;Location 1&amp;gt; with elegant natural movement and product clarity.&lt;/p&gt;

&lt;p&gt;retention_analysis: &amp;lt;Subject 1&amp;gt; (Shot 1): fully_preserved — preserve the facial proportions, center-parted black hair, skin tone and black sleeveless top of &amp;lt;Picture 1&amp;gt;. &amp;lt;Object 1&amp;gt; (Shot 1): fully_preserved — preserve the unusual sweeping chrome top line, dark lens shape and bright polished metallic surface of &amp;lt;Picture 2&amp;gt;; it must rest realistically on her nose and ears. &amp;lt;Location 1&amp;gt; (Shot 1): background_reference — use the circular skylight, pale concrete curves, sunlight geometry and empty architectural volume from &amp;lt;Picture 3&amp;gt;. Do not treat the gray product-photo background as the scene.&lt;/p&gt;

&lt;p&gt;detailed_description: [Shot 1, 00:00–00:05] One continuous photorealistic 85mm portrait shot inside &amp;lt;Location 1&amp;gt;. Medium close-up of &amp;lt;Subject 1&amp;gt; already wearing &amp;lt;Object 1&amp;gt;, standing almost still near a pool of skylight. A very slow camera arc, no more than 20 degrees, moves from her three-quarter left to a more frontal angle. She gently turns her chin toward the light and adjusts the right temple of her sunglasses once with her right hand. The chrome frame catches one clean highlight moving across it; keep the sunglasses design geometrically stable. Realistic hand-to-frame contact, natural eyelid motion, fine skin detail, high-end editorial color. No abrupt cuts or added products.&lt;/p&gt;

&lt;p&gt;overall_soundscape: Open architectural ambience, quiet distant footsteps with gentle reverberation, soft clothing movement and a very faint breeze.&lt;/p&gt;

&lt;p&gt;non_diegetic_music: Three restrained crystalline electronic notes over a warm, minimal ambient pad.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  Setup Configs
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;Chrome eyewear — model, product and architecture Duration: 5 seconds; choose 124 frames / 24 fps. Suggested FIRST test output: about 1 megapixels, preserve reference aspect ratio via the official template if possible. Ref_image_size: match. Steps: 8 | guidance: 1.0 | sigma shifts: video 12.0 / audio 3.0 | VSA: 10%&lt;/p&gt;

&lt;p&gt;Connect ordered inputs into ref_images on MiniMaxH3ReferenceToVideo: &amp;lt;Picture 1&amp;gt; -&amp;gt; Picture_1.png -&amp;gt; the model’s facial identity and dark hair &amp;lt;Picture 2&amp;gt; -&amp;gt; Picture_2.png -&amp;gt; the silver sculptural wraparound frame and dark lenses &amp;lt;Picture 3&amp;gt; -&amp;gt; Picture_3.png -&amp;gt; the circular skylight and sculptural concrete atrium&lt;/p&gt;

&lt;p&gt;Do not connect these images to the First Last Frame to Video node.&lt;/p&gt;

&lt;p&gt;Actual render quality and VRAM requirements will vary.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  Output
&lt;/h3&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/wH4LfL78n0M" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;Generation 3: Character Consistency — 5 seconds&lt;/strong&gt;
&lt;/h2&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 1&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhn9hp28lpbitfpwyoj3d.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fhn9hp28lpbitfpwyoj3d.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 2&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fs95q8qix4do8ai4kd2z4.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fs95q8qix4do8ai4kd2z4.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 3&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftaalmkkw6hnx7v9o3ted.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftaalmkkw6hnx7v9o3ted.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Prompt&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;subject_definitions: &amp;lt;Subject 1&amp;gt; is the fictional woman in &amp;lt;Picture 1&amp;gt;. &amp;lt;Garment 1&amp;gt; is the black oversized puffer jacket with one broad solid red band in &amp;lt;Picture 2&amp;gt;. &amp;lt;Location 1&amp;gt; is the bright circular concrete atrium in &amp;lt;Picture 3&amp;gt;.&lt;/p&gt;

&lt;p&gt;summary: [reference generation] A five-second wardrobe-transfer test: keep the identity of &amp;lt;Subject 1&amp;gt;, dress her in &amp;lt;Garment 1&amp;gt; and place her in &amp;lt;Location 1&amp;gt; in a new shot.&lt;/p&gt;

&lt;p&gt;retention_analysis: &amp;lt;Subject 1&amp;gt; (Shot 1): fully_preserved — use the facial features, long center-parted black hair and skin appearance from &amp;lt;Picture 1&amp;gt;, even though the original outfit and gray studio are not used. &amp;lt;Garment 1&amp;gt; (Shot 1): fully_preserved — transfer the shape, thick quilting, upturned collar, black glossy fabric and red chest stripe from &amp;lt;Picture 2&amp;gt;. &amp;lt;Location 1&amp;gt; (Shot 1): background_reference — match the round skylight, raw concrete and sunlit floor from &amp;lt;Picture 3&amp;gt;. Do not transfer the head or body of a mannequin or an invisible product model.&lt;/p&gt;

&lt;p&gt;detailed_description: [Shot 1, 00:00–00:05] Single unbroken live-action fashion shot. &amp;lt;Subject 1&amp;gt; stands under the skylight in &amp;lt;Location 1&amp;gt; wearing &amp;lt;Garment 1&amp;gt;, zipped up to the collar. Camera begins at waist-height in a medium shot and performs a gentle slow push-in. She looks up briefly toward the skylight, then lowers her eyes to the camera with a subtle serious expression. She slips both hands into the jacket pockets. The red band remains aligned and fixed around the jacket as her posture changes; jacket folds move naturally. Strong architectural shadows, realistic skin and garment materials. No cut, no extra clothing, no text or logos.&lt;/p&gt;

&lt;p&gt;overall_soundscape: Bright quiet hall ambience, a light sleeve rustle, soft echo of distant air circulation.&lt;/p&gt;

&lt;p&gt;non_diegetic_music: Slow minimal analog synthesizer and a low cinematic pulse.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Setup Configs&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;Cross-reference wardrobe transfer Duration: 5 seconds; choose 124 frames / 24 fps. Suggested FIRST test output: about 1 megapixels, preserve reference aspect ratio via the official template if possible. Ref_image_size: match. Steps: 8 | guidance: 1.0 | sigma shifts: video 12.0 / audio 3.0 | VSA: 10%&lt;/p&gt;

&lt;p&gt;Connect ordered inputs into ref_images on MiniMaxH3ReferenceToVideo: &amp;lt;Picture 1&amp;gt; -&amp;gt; Picture_1.png -&amp;gt; face and person identity from collection B &amp;lt;Picture 2&amp;gt; -&amp;gt; Picture_2.png -&amp;gt; jacket and design from collection A &amp;lt;Picture 3&amp;gt; -&amp;gt; Picture_3.png -&amp;gt; new location, different from both source photos&lt;/p&gt;

&lt;p&gt;Do not connect these images to the First Last Frame to Video node.&lt;/p&gt;

&lt;p&gt;Actual render quality and VRAM requirements will vary.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Output&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://www.youtube.com/shorts/tgua6_rCpkk" rel="noopener noreferrer"&gt;https://www.youtube.com/shorts/tgua6_rCpkk&lt;/a&gt;&lt;/p&gt;


&lt;div&gt;
    &lt;iframe src="https://www.youtube.com/embed/tgua6_rCpkk" width="315" height="560"&gt;
    &lt;/iframe&gt;
  &lt;/div&gt;





&lt;h2&gt;
  
  
  &lt;strong&gt;Generation 4: Sci-Fi Cinematic Scene — 10 seconds&lt;/strong&gt;
&lt;/h2&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 1&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7wte5pllofu8w6kha1c0.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F7wte5pllofu8w6kha1c0.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 2&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F14rm3hyff2apbv493ln9.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F14rm3hyff2apbv493ln9.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 3&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fh21c7i9py78901rbmmen.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fh21c7i9py78901rbmmen.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Prompt&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;subject_definitions: &amp;lt;Subject 1&amp;gt; is the fictional adult with short dark hair, a fitted dark long-sleeve top and black cargo trousers in &amp;lt;Picture 1&amp;gt;. &amp;lt;Subject 2&amp;gt; is the fictional adult woman with long center-parted black hair and black sleeveless top in &amp;lt;Picture 2&amp;gt;. &amp;lt;Location 1&amp;gt; is the empty concrete hall with tall windows from &amp;lt;Picture 3&amp;gt;.&lt;/p&gt;

&lt;p&gt;summary: [reference generation] A restrained ten-second science-fiction mystery conversation between &amp;lt;Subject 1&amp;gt; and &amp;lt;Subject 2&amp;gt; in &amp;lt;Location 1&amp;gt;. Each character keeps their own face, hairstyle and outfit.&lt;/p&gt;

&lt;p&gt;retention_analysis: &amp;lt;Subject 1&amp;gt; (Shot 1): fully_preserved — short tousled black hair, angular face, dark fitted top and cargo trousers from &amp;lt;Picture 1&amp;gt;. &amp;lt;Subject 2&amp;gt; (Shot 1): fully_preserved — long center-parted hair, distinct softer face and black sleeveless top from &amp;lt;Picture 2&amp;gt;. Keep the two characters visually distinct; do not blend facial features or swap their voices. &amp;lt;Location 1&amp;gt;: background_reference — retain the empty concrete geometry and diffuse daylight from &amp;lt;Picture 3&amp;gt;.&lt;/p&gt;

&lt;p&gt;detailed_description: [Shot 1, 00:00–00:10] One continuous, steady, cinematic medium two-shot in &amp;lt;Location 1&amp;gt;. &amp;lt;Subject 1&amp;gt; stands on screen left, facing &amp;lt;Subject 2&amp;gt; on screen right, separated by one meter. Slow lateral camera drift. At 00:01 &amp;lt;Subject 1&amp;gt; looks to &amp;lt;Subject 2&amp;gt; and says quietly, [English] Did you hear that? A small pause. &amp;lt;Subject 2&amp;gt; looks toward the upper windows and replies at 00:05, [English] It came from upstairs. They exchange a brief, concerned glance as the camera settles. Subtle expression changes and synchronized lips, natural conversational pacing. Keep both faces visible, correct speaker-to-mouth alignment, no cuts and no other characters.&lt;/p&gt;

&lt;p&gt;overall_soundscape: Two distinct natural speaking voices, long concrete-room reverberation, very distant low mechanical hum, faint wind through the window joints.&lt;/p&gt;

&lt;p&gt;non_diegetic_music: N/A&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Setup Configs&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;Two-character dialogue scene Duration: 10 seconds; choose 243 frames / 24 fps. Suggested FIRST test output: about 1 megapixels, preserve reference aspect ratio via the official template if possible. Ref_image_size: match. Steps: 8 | guidance: 1.0 | sigma shifts: video 12.0 / audio 3.0 | VSA: 10%&lt;/p&gt;

&lt;p&gt;Connect ordered inputs into ref_images on MiniMaxH3ReferenceToVideo: &amp;lt;Picture 1&amp;gt; -&amp;gt; Picture_1.png -&amp;gt; Character A, short dark haircut and black top &amp;lt;Picture 2&amp;gt; -&amp;gt; Picture_2.png -&amp;gt; Character B, long center-parted dark hair &amp;lt;Picture 3&amp;gt; -&amp;gt; Picture_3.png -&amp;gt; quiet concrete hall with windows&lt;/p&gt;

&lt;p&gt;Do not connect these images to the First Last Frame to Video node.&lt;/p&gt;

&lt;p&gt;Actual render quality and VRAM requirements will vary.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Output&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/tS0chhNKt3E" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;Generation 5: Multi-Shot Commercial — 10 seconds&lt;/strong&gt;
&lt;/h2&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 1&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fgvh2d9w47p1di9mw56tt.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fgvh2d9w47p1di9mw56tt.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 2&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F54d4s2ppo0ouwfwjzr4c.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F54d4s2ppo0ouwfwjzr4c.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 3&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftgz2xksvc0xxsh0h8itn.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Ftgz2xksvc0xxsh0h8itn.png" alt=" " width="627" height="836"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 4&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn7w5gdgfrg2qgvihrtu6.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fn7w5gdgfrg2qgvihrtu6.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Input Image 5&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fk0ljlsqb4ra2mjqe2i8i.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2Fk0ljlsqb4ra2mjqe2i8i.png" alt=" " width="661" height="793"&gt;&lt;/a&gt;&lt;/p&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Prompt&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;subject_definitions: &amp;lt;Subject 1&amp;gt; is the fictional adult with short dark tousled hair from &amp;lt;Picture 1&amp;gt;. &amp;lt;Subject 2&amp;gt; is the fictional woman with long center-parted black hair from &amp;lt;Picture 2&amp;gt;. &amp;lt;Product 1&amp;gt; is the oversized black quilted puffer with a bold solid red chest band from &amp;lt;Picture 3&amp;gt;. &amp;lt;Product 2&amp;gt; is the sculptural chrome sunglasses with dark lenses from &amp;lt;Picture 4&amp;gt;. &amp;lt;Location 1&amp;gt; is the circular concrete atrium in &amp;lt;Picture 5&amp;gt;.&lt;/p&gt;

&lt;p&gt;summary: [reference generation] A ten-second three-shot luxury capsule-collection film connecting two models and two hero products in one architectural location. Stylish, restrained, consistent and photorealistic.&lt;/p&gt;

&lt;p&gt;retention_analysis: &amp;lt;Subject 1&amp;gt; appears in Shot 1 and Shot 3: fully_preserved — same short haircut and face as &amp;lt;Picture 1&amp;gt;. &amp;lt;Subject 2&amp;gt; appears in Shot 2 and Shot 3: fully_preserved — same long black hair and face as &amp;lt;Picture 2&amp;gt;. &amp;lt;Product 1&amp;gt; appears in Shots 1 and 3: fully_preserved — exact black puffed silhouette and red horizontal stripe from &amp;lt;Picture 3&amp;gt;. &amp;lt;Product 2&amp;gt; appears in Shots 2 and 3: fully_preserved — stable polished chrome geometry and dark lenses from &amp;lt;Picture 4&amp;gt;. &amp;lt;Location 1&amp;gt; appears in all shots: background_reference — curved concrete, geometric daylight and central skylight from &amp;lt;Picture 5&amp;gt;.&lt;/p&gt;

&lt;p&gt;detailed_description: Cinematic live-action advertising film with clean physical cuts; no superimposed typography or brand marks. [Shot 1, 00:00–00:03.3] Low-angle medium full-body shot of &amp;lt;Subject 1&amp;gt; in the atrium wearing &amp;lt;Product 1&amp;gt;. One confident step, a subtle camera push, and a clean view of the broad red stripe. Sunlight grazes the quilted surface. [Shot 2, 00:03.3–00:06.6] Cut to an elegant close-up of &amp;lt;Subject 2&amp;gt; wearing &amp;lt;Product 2&amp;gt;. She turns her chin a few degrees, catching an architectural reflection on the chrome frame. Same atrium, consistent light. [Shot 3, 00:06.6–00:10] Cut to a symmetrical medium-wide composition of both subjects standing side by side under the skylight: &amp;lt;Subject 1&amp;gt; wearing the red-band jacket and &amp;lt;Subject 2&amp;gt; wearing the chrome glasses. Both face the camera, with minimal natural movement. End on a clean editorial pose. Preserve identities and products across all three cuts, natural hands, no subject merging.&lt;/p&gt;

&lt;p&gt;overall_soundscape: A single slow sneaker footstep in Shot 1, spacious concrete ambience throughout, slight jacket rustle, quiet transition whooshes at the two cuts.&lt;/p&gt;

&lt;p&gt;non_diegetic_music: A minimal atmospheric electronic fashion cue with a low pulse, three soft metallic accents and a smooth ending.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Setup Configs&lt;/strong&gt;
&lt;/h3&gt;

&lt;blockquote&gt;
&lt;p&gt;Three-shot capsule collection mini-commercial Duration: 10 seconds; choose 243 frames / 24 fps. Suggested FIRST test output: about 1 megapixels, preserve reference aspect ratio via the official template if possible. Ref_image_size: match. Steps: 8 | guidance: 1.0 | sigma shifts: video 12.0 / audio 3.0 | VSA: 10%&lt;/p&gt;

&lt;p&gt;Connect ordered inputs into ref_images on MiniMaxH3ReferenceToVideo: &amp;lt;Picture 1&amp;gt; -&amp;gt; Picture_1.png -&amp;gt; first model identity &amp;lt;Picture 2&amp;gt; -&amp;gt; Picture_2.png -&amp;gt; second model identity &amp;lt;Picture 3&amp;gt; -&amp;gt; Picture_3.png -&amp;gt; shared featured puffer jacket &amp;lt;Picture 4&amp;gt; -&amp;gt; Picture_4.png -&amp;gt; featured chrome sunglasses &amp;lt;Picture 5&amp;gt; -&amp;gt; Picture_5.png -&amp;gt; shared location&lt;/p&gt;

&lt;p&gt;Do not connect these images to the First Last Frame to Video node.&lt;/p&gt;

&lt;p&gt;Actual render quality and VRAM requirements will vary.&lt;/p&gt;
&lt;/blockquote&gt;

&lt;h3&gt;
  
  
  &lt;strong&gt;Output&lt;/strong&gt;
&lt;/h3&gt;

&lt;p&gt;  &lt;iframe src="https://www.youtube.com/embed/dmXWdi0whKs" width="710" height="399"&gt;
  &lt;/iframe&gt;
&lt;/p&gt;




&lt;h2&gt;
  
  
  Generations Timings
&lt;/h2&gt;

&lt;p&gt;&lt;a href="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4fli8lt0fvkg65oz55cg.png" class="article-body-image-wrapper"&gt;&lt;img src="https://media2.dev.to/dynamic/image/width=800%2Cheight=%2Cfit=scale-down%2Cgravity=auto%2Cformat=auto/https%3A%2F%2Fdev-to-uploads.s3.us-east-2.amazonaws.com%2Fuploads%2Farticles%2F4fli8lt0fvkg65oz55cg.png" alt=" " width="702" height="788"&gt;&lt;/a&gt;&lt;/p&gt;




&lt;h2&gt;
  
  
  Conclusion
&lt;/h2&gt;

&lt;p&gt;These experiments show why reference-based video generation is worth testing beyond simple image-to-video prompts. FastH3 Omni gives us a way to combine several visual references and direct how they should appear in the final scene, but the results still depend on the prompt, reference images, and generation settings.&lt;/p&gt;

&lt;p&gt;The most useful part of this process was comparing the outputs across different tasks instead of relying on one successful clip. If you're experimenting with AI video in ComfyUI, the prompts and configurations here should give you a practical starting point for your own tests. I'd be interested to see how the results change with different reference images, longer sequences, and hardware configurations.&lt;/p&gt;




&lt;h2&gt;
  
  
  &lt;strong&gt;FastVideo FastH3 Omni on&lt;/strong&gt; &lt;a href="http://HexGrid.cloud" rel="noopener noreferrer"&gt;&lt;strong&gt;&lt;u&gt;HexGrid.cloud&lt;/u&gt;&lt;/strong&gt;&lt;/a&gt;
&lt;/h2&gt;

&lt;blockquote&gt;
&lt;p&gt;&lt;strong&gt;&lt;em&gt;Run FastVideo FastH3 Omni on HexGrid in one click — no CUDA setup, no dependency hunting, no environment debugging.&lt;/em&gt;&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;The template comes preconfigured with ComfyUI, the required runtime, and the&lt;/em&gt; &lt;strong&gt;&lt;em&gt;FastVideo FastH3 Omni&lt;/em&gt;&lt;/strong&gt; &lt;em&gt;setup so you can launch a GPU workspace and start generating immediately.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;strong&gt;&lt;em&gt;No manual CUDA installation. No PyTorch version matching. No broken requirements. No hours spent wiring the environment together.&lt;/em&gt;&lt;/strong&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Just choose the&lt;/em&gt; &lt;strong&gt;&lt;em&gt;FastVideo FastH3 Omni&lt;/em&gt;&lt;/strong&gt; &lt;em&gt;template, start the instance, and open ComfyUI.&lt;/em&gt;&lt;/p&gt;

&lt;p&gt;&lt;em&gt;Pricing starts from&lt;/em&gt; &lt;strong&gt;&lt;em&gt;$0.58/hour&lt;/em&gt;&lt;/strong&gt;*, so you can spin up a GPU only when you need it and shut it down when you’re done.*&lt;/p&gt;
&lt;/blockquote&gt;

</description>
      <category>fasth3</category>
      <category>comfyui</category>
      <category>aivideocreation</category>
    </item>
  </channel>
</rss>
