I’ve recently created my own picture creation app powered by the DALLE-2 API. But to be honest the quality of the generated pictures are really not good.
So, how do i came to this statement.
Microsoft Bing Image Creator is also powered by DALLE-2. When i compare the results of an exact same prompt from my API DALLE-2 app to Bing Image Creator then i have to say there are worlds between the quality.
The Bing pictures are much more better than the API results.
So, the question is, am I doing anything wrong? Do i have to add something in the prompt? In API developer documentation i couldn’t find anything.
a dog running in a garden, black and white, high quality, high resolution, clear image, white background, ad ink outlines
With Bing Image Creator → exact the result i expected
With the API → the image is blurred and has no clear lines
Maybe someone got the same results and have an idea if there is something i can do about and how.
Thank you very much and Happy Easter
I think you would need to look into doing some model training on specific image sets, or possible use it with a GPT4 plugin.
I do not have much experience with images but seen some info in youtube videos.
Bing may be using GPT-4 or a higher version.
Hey Oliver, I had noticed the same things. Images seem to be far lower quality at the moment using the API than I have had previously. As you did, I have also done some tests against the Bing image creator - and, while I appreciate it might include some newer capabilities - the results are drastically different through all of my tests. It is rather strange. I feel better knowing someone else might be seeing the same thing so it’s not only in my head
Ah, thank you for this reply. I thought i was the only one. Hm, how could we get some insights from OpenAi team
I would love to use Dalle-2 for my application. But with the current quality it is not interesting for me.
I’m noticing a world of difference too, between API vs Bing and API vs Dall-E web app by Open AI. Both Bing and Dall-E web app are superior in quality. All same prompts at 1024x1024 on API:
A photorealistic full-body, front view, product shot of a backpack with teddy bear design, front pockets, brown , professional design, 4k, white background
here’s the API:
Yes same here what we notice that the quality of the output in DALL-E Web UI is better than the API. It is possible that they use more light weight model on the API, but I would say the quality should not be impacted.
Do you know how we can escalate this question?
Yes, we have also noticed this. I personally suppose this is a pure intention, and it might be the case they even lowered the produced quality of images (at the end of last year, in the beginning of this year). For example to reduce computation time (on server side) an to prepare charging maybe more money in the future for more specialized services (instead of getting everything with a flat rate).
If you want to build something using the OpenAI API for DALLE2 you will get very often inferior results in comparison to Bing and especially Midjourney at the moment.
Same here, very simple and poor results compared with bing and midjourney. Im paying for API usage, but the results are not enough to public my app.
Any updates on this?
And same here. The API generated results are quite disappointing. Even if they say this is Dall-e 2, I have my doubts.
I am willing to pay more to get good results rather than paying 2 cents for a picture I cannot use at all.
Let ne know if there is a new gateway. Currently I use v1: POST https://api.openai.com/v1/images/generations
OpenAI have a pretty regular pricing adjustment seen, where they price relative to compute resources. This reveals, for example, how much less processor time is used by gpt-3.5-turbo than davinci, or how much more by GPT-4.
I think you’ve identified the relationship here. 1024px images via API = $0.02, while through labs.openai.com you must prepay $15 for $0.13 images. One of these isn’t a secret backdoor to obtaining the identical thing at reduced prices.
I would not mind paying a higher price to get a good result. They could provide a different access point for high-end pictures, and users can decide what quality they want.
But not giving this choice is not sufficient. I now check stability.ai instead, like others probably do as well. And as soon as midjourney provides API access, latest then OpenAI must react.
BTW, to me it seems the Dall-e API pictures get worse day by day.
I don’t believe we are getting Dallee-2 on the API. We have tested this a LOT and the online version of Dallee gives far better images.
I tested now the stability.ai API, and the results are satisfactory for me. It is not as fast as Dall-e, but for me this is not a critical factor.
I do hope that anyone in OpenAI reads this chat. Please feel free to join in.
Yep I’ve noticed this for months.
I had this big idea that I could have an electronic photo frame in my house and update it daily with cool Dall-E images using a random prompt generator. I’ve written the software and currently use it for my laptop background, it updates with a new image daily and it works well, but the image result are frankly terrible. They look like they were made with crayons by a 4 year old.
I put the exact same prompt into any online AI image generator and the results are far superior every time.
I’m paying to use this API and the results are shocking.
Surely the OpenAI devs are aware of this??!
Same idea I had DiWa project #33: AI weather forecast display frame with DALL-E - YouTube
I combine the random objects with the weather forecast on my digital frame.