Create AI Images : Image to image

-

 Image generated using the Z-image Turbo model (image to image) πŸ”»

Image generated using the Z-image Turbo model (image to image)


In my previous posts, we explored how to utilize Text-to-Image (txt2img) on lower-end PC setups and how various parameters influence the final output. Today, we’re shifting our focus to a more dynamic tool: Image-to-Image (img2img).

Using the Z-Image-Turbo (ZIT) model once again, I’ll show you how to achieve stunning results without needing a high-end workstation. While results vary based on individual settings, this guide focuses on generating images that stay faithful to the original composition while enhancing quality and style.


Workflow & LoRA Setup

To follow along, you’ll need the custom workflow and the specific LoRA model I’ve prepared. (Note: A LoRA(Low Rank Adaptation) is a specialized tool that allows the AI to generate specific styles—like pixel art or oil painting—with high efficiency and precision.)

  1. Download: Access the files via Download Workflow & LoRA Here.

  2. Installation: Place the downloaded LoRA file (pixel_art_style_z_image_turbo.safetensors) into the ComfyUI/models/loras directory.

  3. Launch: Open ComfyUI and press Ctrl + O to load the workflow(json) file.


The Workflow: Transforming Your Images

Once the workflow is loaded, you will see the following layout:

layout

  • Step 1: Locate the Load Image node (highlighted in the red box) and upload the image you wish to transform. For this demonstration, I used an image of Shorekeeper from Wuthering Waves.

layout step1

  • Step 2: Enter your prompt. For this demonstration, I used a prompt designed to maintain a realistic aesthetic without drifting too far from the original photo.

step2

  • Step 3: Queue the prompt. After a brief processing period, you’ll see the transformation.

step3

 The result is surprisingly crisp and aesthetically pleasing.


Controlling Creativity via Denoising & CFG

The true power of img2img lies in fine-tuning your parameters. While you can experiment with many settings, two are crucial for ZIT:

  • CFG (Classifier Free Guidance): For this model, I recommend keeping this between 1.0 and 2.0. It keeps the generation stable and aligned with the prompt.

  • Denoising Strength: This is your primary lever for "creativity." Think of Denoising as "how much of the original image should the AI erase and redraw?"

  1. Low Denoising (0.3): The AI stays very close to the original structure, making subtle refinements.

    Low Denoising

  2. High Denoising (0.9): The AI treats the original image as a loose sketch, generating something almost entirely new.

    High Denoising

Wrapping

Today, we’ve only scratched the surface of what’s possible with img2img. By customizing your workflows or switching models, the possibilities are endless—from forcing specific poses to maintaining character consistency across different actions, or even training your own custom models. 

In the next post, we will dive deeper into Pose Control and Custom Model Training. 

Comments

Post a Comment

Popular posts from this blog

Training Custom LoRA for Z-Image-Turbo (ZIT) via Local PC

-
Image
⚠️ Important Disclaimer & Ethical Use Before proceeding with the tutorial, please acknowledge the following ethical guidelines regarding AI model training: Non-Real Person: The character trained and demonstrated in this guide ( Nano-Banana ) is a purely fictional AI-generated character and does not represent any real-life individual. Responsible Use: I strictly discourage and prohibit the use of this technology to recreate real people without their explicit consent. Please use these tools ethically to respect the privacy and rights of others. ←Prev Post Image generated using the Z-image Turbo model (Generated LoRA ) πŸ”» Introduction In this post, I will provide a comprehensive guide on training a custom LoRA to achieve character consistency using the Z-Image-Turbo (ZIT) architecture. It is important to note from the outset: LoRA training is a resource-intensive process that demands significant h...
Read more

Create AI Images : High-Speed Pose Control & Custom Character LoRA Workflow for ZiT

-
Image
←Prev Post   Image generated using the Z-image Turbo model (LoRA -  L1n4 ) πŸ”» In the world of AI image generation, speed is no longer enough—we need control. In this post, we’ll explore how to harness the power of Z-Image-Turbo (ZIT) to achieve two critical goals: capturing precise physical postures through Pose Control and maintaining consistent character identity using Custom LoRAs. Whether you're aiming for a specific action or a recurring character, this workflow will give you the steering wheel. 1. Dynamic Pose Control with ZIT ZIT is remarkably efficient, and when combined with ControlNet, it allows for near-instant generation with exact spatial accuracy.  In general, image generation models tend to drift toward common or “average” poses rather than strictly preserving a given one. This workflow addresses that limitation by continuously anchoring the pose through ControlNet, allowing the model to maintain the intended spatial structure with high consistency. Access...
Read more

Creating AI Videos on an RTX 4060 (LTX-2 vs. Wan2)

-
Image
  Image generated using the LTX-2 model πŸ”» Until now, we’ve focused on generating high-quality still images. However, in today's content landscape, static images have their limits. To push the boundaries of AI creativity, it's time to dive into AI Video Generation . Generating video is a completely different beast compared to images. Video models are massive, requiring significantly more storage and high-end hardware. If your system barely manages low-spec optimized models like ZIT (Z-Image-Turbo) , jumping into video might feel like hitting a wall. The Reality Check: Wan vs. LTX When I tested Wan2  on my RTX 4060 (8GB VRAM) , it was a struggle. The generation took forever, and I faced constant "Out of Memory" (OOM) errors. On the other hand, LTX-Video managed to run relatively well. It took about 10 minutes to produce a 5-second clip . For a local 4060 setup, this is a massive win. Getting Started...
Read more