Hey everyone, I've hit a real head-scratcher with the latest `v6.1` model. For years, the `--no` parameter has been my go-to tool for refining a prompt by explicitly telling the AI what *not* to include. It's fundamental to my workflow for generating clean marketing assets.
Lately, however, it seems completely broken. I'm getting the exact elements I'm trying to exclude, which is ruining batch jobs and eating through fast hours. For example, this prompt used to work flawlessly:
```
/imagine prompt: a modern tech logo, sleek and minimalist --no text, words, letters, taglines, symbols
```
In `v6.1`, I consistently get a logo with stylized text or embedded letters in the design. The negation is being ignored. I've tried all the troubleshooting steps I know:
* Using `--no` at the end of the prompt, as per usual.
* Trying different weightings like `--no text::2`.
* Placing the `--no` directive right after the main subject.
* Using simpler, single-word negations (`--no text`).
Nothing works. It feels like the parameter is just parsed and then discarded.
Has anyone else run into this? I'm wondering if:
* This is a known bug in `v6.1` that's being addressed.
* The syntax or weighting system for negations has changed silently.
* There's a new, undocumented way to achieve the same exclusion effect.
Any workarounds or insights would be incredibly helpful. My current Zapier automation that relies on this is essentially on hold until I can figure it out.
api first
api first
This is rough, sorry you're dealing with it. I'm actually in the middle of evaluating whether to migrate some of our legacy content generation workflows over to this platform, and seeing something like the --no parameter just stop working is kind of terrifying. If I can't trust basic exclusion logic, how do I even plan a timeline for moving our asset pipeline? I'd need to budget for manual cleanup or extra iterations, and my boss wants a hard estimate.
Have you checked if the weighting syntax still works in other ways, like using negative weights on certain keywords in the main prompt itself? Or is that also broken? I'm trying to figure out if there's any workaround at all, because if not, I might have to stick with our old system until 6.2.
One step at a time