I've been using the Drawing Assist feature on the Samsung Galaxy S25 Ultra, and while it has potential, the restrictive filters and lack of flexibility make it frustrating to use. Creating anything satisfying requires constant tweaking or spamming the "generate" button, only to be met with messages like "Can't generate because the prompt might contain inappropriate content" or "Can't generate with this content."
Some of the filtered words and phrases make no sense:
Filtered word: Chest
Prompt: Toy chest
Filtered word: Magic wand
Prompt: Wizard with magic wand
Filtered word: **bleep**
Prompt: Deviled eggs
Filtered word: Hoe
Prompt: Garden hoe
Filtered word: Slime
Prompt: Snail leaving a slime trail
There's no clear reason for these restrictions other than an overly strict word filter.
Beyond that, the tool struggles with specific prompts. For example, asking it to generate a wrecked pirate ship results in a standard pirate ship with no visible damage. It either wasnāt trained on that data or is outright refusing to generate it. Similarly, requesting a demon doesnāt flag a warning but instead generates a dragon. And when it comes to humans or humanoid figures, it refuses entirely; even something as simple as an elf in a fantasy setting.
If Samsung plans to start charging for this feature by the end of the year, these issues need serious attention. The restrictive filters need to be loosened, and the tool requires a broader training dataset to handle more prompts accurately. As it stands, the limitations make it more of a hassle than a creative assistant.
Additionally, more features need to be added; such as more aspect ratio options and the ability to change orientation (e.g., 1:1, 4:3, 16:9, 3:4, 9:16). If this is meant to be a professional-grade tool, these issues need to be addressed before monetization; otherwise, it risks frustrating users rather than helping them create anything worthwhile.
