System Requirements

We recommend Google Chrome, but any modern, updated browser should work fine. If you have antivirus software or browser extensions, please consider whitelisting the website Kenna.ai to avoid display issues, as those can sometimes block harmless JavaScript tags used in our User Interface.

You can use it from your smartphone, tablet, or computer. Most computers built in the last 10 years can run Kenna AI, regardless of brand and Operating System.  

Blurry Images

The first image that appears is a draft, meaning you can take the quality much further by moving it to the Upscale tool. Also try adding hashtags like #kennaxl as explained in the prompting advice section.

Blue Artifacts

99% of the time, the Guidance is set too high when too many specific or conflicting words are in a prompt. To fix it, try re-running the image at a lower guidance. If that doesn’t solve the issue, please try and reducing the number of models or positive/negative prompts

Duplicate subjects 

This is happening when the AI model doesn’t know how to fill in the blanks. To fix it, try reducing the resolution of the Generate image target lower than 768×768 and start the prompt with “one object, one woman, one man” and fill in the background details by describing them.

Cannot get multiple subjects right

Kenna’s models are product-focused, so the tool greatly prefers to focus in on a single object, but this is of course possible. There are a ton of different ways to go about this, and here are some examples:

  • Make a rough sketch of everything that you want to see in your photo, and then use the Sketch-to-Image tool to bring those things into realistic focus
  • Use the Generative Fill tool to add subjects one by one with accuracy. Also check our guide on word order, it can make a big difference. 
  • Instead of using the Generate tool, find a similar photo or composition online. Switch to Create Variations tool to replace the image with what you’re describing
  • Beta Feature: If you’re writing a very long prompt, try adding /lpw in the positive prompt box (experimental feature, long prompt weights).  This parameter will change the way the A.I. balances each word, with the tradeoff that it may weaken your negatives to compensate.

The image quality is blurry or washed out

Check the resolution. The draft resolution is 512×512, so try 768×768 or upscaling the image instead.  Prompting “sharp focus” or “macro photo, close-up” and such in the positive field can improve the image. Likewise, use the negative field to put “blurry, bokeh, out of focus” to remove such pictures from the possible outcome.

The image is too zoomed in

Certain keywords like “portrait” will affect the image zoom and also background focus. Portrait is typically used when the person is close to the camera, feet cut off, with a soft or heavy bokeh background. Consider using other camera shot terminology.

The feet are cut off, cannot get a full body image

The easiest way to solve this is to describe the shoes or feet towards the front o the prompt, and to choose an aspect ratio that would lend best to this pose.

The faces or hands don’t look right

This one’s a bug impacting the whole AI industry. Computers are especially bad at figuring out what to do with our hands. There is hand-corrective technology running in the background on every image, and we are working to improve it.

Try adding /adetailer to your prompt.  This will search the image for and faces hands, and replace them at run time.  Add this to the positive prompt field:

/adetailer

Other methods: If the image is only a little off, sometimes the hands and faces will correct themselves when requesting more images. Try generating more Variations and an image with good hands will eventually appear (some images are tougher than others) or use the Generative Fill tool to draw over the hand. When all else fails, try a different gesture.

Account Issues

Forgot password

Click the Get Started button and follow the steps to reset your password. If you don’t receive it right away, please check your spam folder or try a different email address. Please check the email address on the copy of the welcome email or receipt to be sure. If you’re locked out please contact us.

Billing Issues or Account Changes

Downgrading, changing plans, refunds or canceling accounts please contact us.

Limitations

Limited knowledge before 2021

Kenna AI was trained in part with billions of images from the public internet before 2021.  As such, some pop culture references may not yet be understood. Please discuss and request any blind spots in our Discord community.

 

The Generate tool is different from a chat bot

Unlike a tool like Chat GPT, when we are using the Generate Tool, we are not “chatting” with Kenna AI. Every image is a fresh start, so write the full image description every time. 

The Generate Tool does not understand prompts like “make a change” or “what if it was like this” commands, , because it isn’t talking at all. Generate is only creating what is in the prompt field. 

When it’s time to make an edit, rephrase the original prompt or for spot changes try the Variations and Generative Fill tools instead.