Did I stumble upon a new feature or did I miss a recent announcement about this?
The prompt’s custom instruction needs some adjusting but… this is a game changer! I’ll provide some feedback about it below the screenshots.
Some feedback:
- There should be a way to set the name of the asset when it’s saved. Right now, it saves with the prompt as the asset name.
**Edit: ** Oh, I just realized that the save filename is right above the Save button. That wasn’t obvious to be at first. There should be a “Name” label right above it.
Also, I was able to save both generated images with the same exact name. That should be prevented and a unique number should be appended to the filename (e.g. alien, alien 1, alien 2, etc.).
Otherwise, this happens:
And a bug occurs when I select either of the two filenames in the screenshot below. Only one of the image assets is displayed. There is no way to select the other image.
- I wonder which AI API it’s using. DALL•E?
- The generated image is not zoomed properly:
-
This is not specific to AI image generation but in the list of assets under Media Files, there should be a way to mouseover or click on an image name and view the asset without assigning it to an image component on a screen. This is especially true since the zoom level isn’t correct on generated images.
-
The Inline Assistant (a setting in the AI section of the Design tab) provides a way to generate images from prompts directly into an image component on the Design tab but it doesn’t work yet. There is no way to enter text into the prompt field.
- There needs to be a way to crop generated images. Otherwise, you can get this problem where the generated image has too much white space around it. And since there’s no simple way to export the generated image, I can’t just use remove.bg. Edit: I was able to drag the generated image from the generator panel to my desktop.
And since we’re on the topic of AI and Thunkable already makes use of Cloudinary, why not offer users a quick way to remove the background and crop image assets?