Skip to main content

Image Gen V2

Feature Release: Image Generation - Generating images, now simplified.

Ayumi Umehara avatar
Written by Ayumi Umehara
Updated over a week ago

Overview

Introducing the new Image Gen V2, an overhaul of our legacy image generation interface, designed to simplify the image generation process for all users. With Image Gen V2 comes new presets designed to aid you in achieving specific aesthetics with less effort.
​


About Image Gen V2

  • ​Presets: Image Gen V2 reduces the complexity of the image generation process with the introduction of Presets, which includes pre-defined combinations of models, styles and Elements and other parameters. These help users effortlessly achieve specific aesthetics while reducing the need for complex prompting.
    ​
    In addition, you will be able to create and save your own presets for even more personalized workflows. (Coming Soon)

  • Familiar tools: Image Gen V2 is an simplified experience but also has the same features users know and love integrated.


    These include:

    • Elements

    • Image Guidance

    • PhotoReal

    • Transparency
      ​

  • Streamlined workflows: Image Gen V2 simplifies the process of using Style Reference and Character Consistency placing it right into the prompt bar. Alongside that comes a new Image Guidance menu to better aid the user in selecting the mode they require.

  • The best of the best: To offer the best quality generations possible, Image Gen V2 uses only Phoenix, Flux and SDXL based models, which allows for higher resolution imagery and utilizes our best workflows.
    ​

  • Contrast: Image Gen V2 comes with a new Contrast setting, allowing you to choose between 3 different contrast options (Low-High).
    ​

  • Image Dimensions: Pick from several widely used resolution options. Alternatively you can choose how wide or tall your image should be and we will automatically optimize the resolution for the best output based on the model.

Benefits

  • Workflows simplified: Generating images with consistent characters or restyling an image has never been simpler. We have simplified and made Style, Character and Content Reference much more easily accessible.

  • Clean UI: Collapsable sidebars, style presets and simplified settings. Image Gen V2 allows you to focus on your images while keeping other things out of the way.

Settings

Settings for generating images can be found on the left sidebar. It is advisable to try out the various settings to get familiarized with the different options available to get the most out of the new Image Generation tool.
​
​

  • Generation Tools:

    Switch between the various generation tools available within the AI Creation page, including image generation, Video generation and Flow State.
    ​
    ​
    ​

  • Preset:
    ​


    This displays the current active preset, clicking on it will open the Presets library and display additional image options where you can select from one of several available presets and change settings.
    ​

  • Prompt Enhance (Phoenix and Flux only):
    ​


    Automatically enhances short prompts by default on Auto. Can be switched to On to always enhance all prompts regardless of length or completely disabled.
    ​
    It is recommended to keep it active when using Phoenix and Flux for optimal results.
    ​
    ​

  • Style:
    ​
    ​
    ​
    Each preset comes with an selection of style presets which are tailored to various categories of imagery.

  • Contrast (Not available for Flux):
    ​
    ​
    ​
    Give your images intense contrast by changing the Contrast setting to High, or make things less harsh by switching it to Low.
    ​

  • Generation Mode (Not available for Flux):
    ​
    ​
    ​
    Pick between Fast or Quality.

    • Fast: Reduced generation time at the expense of quality and resolution limits. Not optimal for extreme aspect ratios.

    • Quality: Slower generation times, however image detail and quality is drastically improved. Excellent for extreme aspect ratios and higher resolution images.

    • Ultra (Premium Only): High resolution images with upscaled detail. Only available with Phoenix.
      ​

  • Image Dimensions: Pick from 3 options in the sidebar or click on More to view additional Aspect Ratio and Custom Resolution options.
    ​


    ​

  • Number of Images:
    ​
    ​
    ​
    Select the number of images you would like to generate.
    ​

  • Privacy (Premium only):
    ​
    ​
    ​
    Enable this option to generate images privately without them showing up in the public feed.
    ​

  • Advanced Settings:
    ​
    ​
    ​
    Select the Model used to generate images, or toggle several options -

    • PhotoReal - Enhances image aesthetics further. (Compatible with SDXL models only)

    • Negative Prompt (Compatible with Phoenix and SDXL models only)

    • Transparency - Generates images with alpha transparency (Compatible with SDXL models only)

    • Tiling (Compatible with Phoenix and SDXL models only)

    • Use Fixed Seed - This can be used to regenerate an generated image (if a similar prompt and settings is used). The seed is essentially an starting point from which the image is generated from.

In addition, Image Guidance (Style, Character, Content Reference etc.) and Elements (including custom trained Elements) can be found in the button in the left side of the prompt input.

Image Guidance: These allow you to use an reference image to influence the generated image in several ways. Please consult the Image Guidance guide for additional details on the various avaliable options.

Elements (LoRAs): These allow you to influence the style of the image. Up to 4 Elements can be mixed and matched to generate images of a unique style. Please consult the Elements guide for additional details.


In addition custom Elements for Flux and SDXL models can be trained to apply your own style to the image or generate an consistent character or object in various situations. Please consult the Element Training guide for additional details.
​

You may also generate an Random Prompt, improve your existing prompt, edit the prompt, use AI to generate an prompt with an image by clicking on the button to the left of the Generate button and selecting the appropriate option.
​

Describe with AI

A useful feature that can be helpful in identifying visual elements such as a print style or fashion look in a existing image or just generally recreating an image that may resemble the sample image, Describe with AI essentially examines an provided image and suggests an prompt based on the image.

To use this feature, simply click on the button to the left of the Generate button and select Describe with AI, then upload or chose from an existing image. A generated prompt will appear after, which can be further modified if needed.

⚠️ Note:
​
As Describe with AI does not reverse engineer the original prompt used by a generated image nor select the best matching preset, model and Elements, it may be unable to recreate images exactly.
​
Describe with AI may also be unable to properly recreate complex scenes, such as locations in specific cities nor recreate characters. In such cases, we recommend training an Element on them instead.

Using Image Gen V2

Using Image Gen V2 is a fairly straightforward process if you are new to image generation.

  1. Initiate: On the main app, click on Image in the left sidebar.
    ​

  2. Stylize: Select the Preset and Style that best suits the image you have in mind. (Note: You can scroll up/down the Style menu to view more available styles for the current preset.)
    ​

    ​

  3. Size-up: Select your preferred Image Dimension and Number of Images.
    ​

  4. Prompt: Enter your prompt, ensuring that you describe your image in visual terms as much as possible. Short prompts are excellent if you are just looking to generate an simple image.
    ​

  5. Generate: Click on the Generate button to start the image generation process. Note that the button will display the cost of the generation job. (Refer to the counter on the center top of the page for your total token allowance)
    ​

    ​
    ​

Image Options

Hovering over an generated image reveals several quick actions.
​

  1. Multi-select: Allows you to select specific images to apply collective actions to, such as changing image visibility, adding to an collection or downloading.

  2. Collections: Organize your generations into Collections for better curation.

  3. Download Image: Saves your image onto your computer

  4. Make Image Public / Private: Changes the visibility of an image to other users.

  5. Upscale: Increase the resolution and quality of your images through various upscale options including Ultra.

  6. Edit in Canvas: Brings your image into Canvas Editor for further edits such as inpainting to modify specific details of the image or outpainting to extend the image.

  7. Generate Video: Turn your images to short videos easily.

  8. More options: Delete the image, copy to clipboard, remove background etc.


    ​

You may also click on the … button next to the image details to apply certain actions to an image batch including deleting the entire batch.

Frequently Asked Questions

Is Image Gen V2 available for free users?

  • Image Gen V2 is available to all users.

What are the token costs for generating images?

  • The token costs for generating images are highly variable. The end cost is affected by several factors which adds up to the total amount:

    • Model used

    • Generation Mode

    • Resolution

    • Number of Images

    • Image Guidance options

    • Elements (The cost is same regardless of amount of Elements active)

​

The model I used to use is unavailable in Image Gen V2

  • As Image Gen V2 only utilizes Phoenix, Flux and SDXL models, you will have to switch back to Legacy Mode, which can be enabled on the top right corner of the page.
    ​

The quality of my images are poor

  • This could be due to several different factors. However you may do some checks on the follow to ensure your settings are as optimal as possible:

    • Reducing the strength of the Elements - Some models may require that the strength of Elements be reduced to get the best images possible.

    • Using Quality for the Generation Mode - This vastly improves the image quality especially for higher resolutions.

    • Increasing the Image Size - Some models may work better with larger Image Sizes. Typically it is best to use Medium as a default.

    • Reduce the Contrast setting - High contrast setting may result in unwanted excessive darkening of your generated images.

    • Check if the Use Fixed Seed option was enabled.

    • Check if any unwanted Image Guidance options are active.

Which preset / model should I choose?

  • Generally as most use cases vary, there is not one model that is a one-size fits all option. However here are some tips to help you pick the model that may be best suited to your requirements. As a general tip, it is always good to experiment with the various presets and models available to discover what works best for you.
    ​

  • For great aesthetics and short prompts and simple scenes: Models such as Kino XL and Vision XL. (Try presets such as Cinematic Kino, Lifelike Vision and for stylized images, Anime).

  • For varied aesthetics including stylization with complex scenes or text: Phoenix is optimal for various aesthetics and text generation alongside with great prompt adherence, allowing for more complex scenes.

  • For highly photorealistic and detailed prompts or pairing with custom Elements: Flux Dev is recommended. It follows detailed prompts well and has great photorealistic aesthetics.

Feedback

As Image Gen V2 is an entirely new experience, we value your feedback and thoughts on how we can improve the generative process for our users through this feature.

Please feel free to leave your feedback here.

Attachment icon
Guidance_Elements_.mp4
Attachment icon
Describe_AI_gif.mp4
Did this answer your question?