Unpredictability of Image Generator output format
一位用户对图像生成器感到满意，但对其不可预测的输出格式感到沮丧。输出格式可能是一个单独的图像、两个供选择的图像或四个备选图像。用户指出，目前没有明确的方法来请求特定的输出格式，并且尝试通过要求“想要四个图像：1. (blah), 2. (blah)等”来鼓励生成四个图像的尝试常常被忽略，或者导致四个备选图像出现在一个图像中。
时间与来源
时间显示为 UTC
显示时区：UTC
本地时区尚不可用，暂时显示 UTC。
发布当时偏移：UTC+02026年9月11日 18:58 UTC
收录当时偏移：UTC+02026年9月12日 12:01 UTC
- 发布
- 2026年9月11日 18:58
- 收录
- 2026年9月12日 12:01
- 来源类型
- 开发者社区
- 档位
- 社区
- 信源状态
- 正常
档位是按信源手工设定的编辑判断，不是逐条打分。
Hey all, I'm really enjoying the image generator, but I'm a bit frustrated over the lack of control over how the images are presented. I might be given: A single image Two images that I'm asked to choose between (I guess for training) Four alternate images
Sometimes (one time in fifty?) I'm given a "Preview" image before the final post processing. Often I prefer the preview image, and I can save this before GPT replaces it with the final image.
I am more likely to get preview images, if it also decides to give me a carousel of four images.
As far as I can tell, there is no way to specifically ask for the output format. I _think_ I can encourage it to produce four images if I ask " want four images: 1. (blah), 2. (blah), etc", but it often ignores this, or places four alternatives into a single image (which is okay, I guess...)
Prompting "please provide previews" doesn't do anything.
I do appreciate that the image generator and the chat system are two separate system (as Chat takes pains to inform me.. "I have no control over what the image generator does", so there is this wall that can't be bypassed), but has anyone found any other ways to encourage the generator to produce certain output formats?