Hello all,

I thought there would be a pinned post for this but alas no

Having recently got into AI myself I have tried various sites to generate WAM content...some sites like Sora generate a lot of content warnings (even with very basic prompts)
and I have tried looking through this Forum to see which sites people are using to create their content...it's not easy, as most people don't credit the site they used...

So I am wondering, which sites have you found are best for?:

- Creating Great WAM images, which feel real
- Excellent WAM videos that really capture the physics and dynamics
- Which sites can be used for NSFW content? i.e. Nudity with WAM

I would ask the creators out there to share their best experiences, I am happy to subscribe to a good site...but these places often aren't cheap, so I'm hoping to leverage from your experience!

My ideal goal would be to create a long video of my own (obviously clipping together the smaller videos that AI can generate)
Ideally with sounds and voice

So please, help out a budding creator and others who want to explore AI creation

Or if I have missed a thread on this, please point it out to me!

There really isn't much choice out there.

For images,
OpenAI
ChatGPT4o/Sora.

Google
ImageFX, AIStudio, Gemini, Whisk.

Most other sites are really pretty poor.

Video.

Google's VEO3 is the best out there. You'll struggle with Sora for most part. hailuo and kling aren't too bad either but will need more work to get your prompts working.
If you want sound and voice, VEO3 is your only option atm.

Thanks for the feedback, good to know there are limitations out there...seems a lot of people are doing more advanced stuff with training models. I've been dabbling on some AI girlfriend sites, but results are often poor on their generators

I've been using the Google suite of AI tools, as I already have a workspace account with them (I'm the only "employee" though.)

Whisk with Imagen3 is great for images, as well as Veo3 via Gemini for video. Compared to other AI tools, it allows more "adult" content, but likely not nudity or anything overtly sexual, which works for me

Examples attached - I just have an animated GIF of the video, but the video does have sound:

As people said, when it comes to closed-source models it's either the GPT one available in ChatGPT / SORA or the Imagen model available in ImageFX or Whisk.

Which is better depends on what you want to do and how you have optimized for specific prompts.

IMO highest quality is the image generation tool of ChatGPT, especially used with custom instructions. But it can be very picky about content filtering and you really need to massage the bot the right way.

ImageFX (after its recent update) is best for model creation and lighting but it still struggles with complex scenes or physics. On the other hand it has very loose content filtering compared to the other options.

None of the closed source models will let any NSFW content pass through at this point, image recognition has simply become too good.

I can't really say anything about video, in my personal and humble opinion nothing is really good enough at the moment and it might need another model generation or two before it changes.

It's not the most sophisticated but Perchance will allow NSFW if you tell it.

Except that it's terrible at messy images. Unfortunately you need an auto regressive model like GPT4o/Sora or Imagen4 to do anything remotely messy.

The alternative and probably the best option is to train your own model or use a specifically trained lora but that's not particularly accessible to most people.