Basictek

Just thought I'd share some of my experiments with LLM/AIs to make some ASFR and Freeze-adjacent videos featuring larger women.
https://www.redgifs.com/users/haaaardtimes

Great stuff! Been following you on DA for while, thank you for posting these!

as someone who loves seeing bigger women frozen…i loved this!

ensignronald wrote: Fri May 30, 2025 11:23 am Just thought I'd share some of my experiments with LLM/AIs to make some ASFR and Freeze-adjacent videos featuring larger women.
https://www.redgifs.com/users/haaaardtimes

what AI did you use to make videos?

neisy03 wrote: Fri May 30, 2025 9:41 pm

ensignronald wrote: Fri May 30, 2025 11:23 am Just thought I'd share some of my experiments with LLM/AIs to make some ASFR and Freeze-adjacent videos featuring larger women.
https://www.redgifs.com/users/haaaardtimes

what AI did you use to make videos?

Kling.ai for almost all of them.

ensignronald wrote: Fri May 30, 2025 9:48 pm

neisy03 wrote: Fri May 30, 2025 9:41 pm

ensignronald wrote: Fri May 30, 2025 11:23 am Just thought I'd share some of my experiments with LLM/AIs to make some ASFR and Freeze-adjacent videos featuring larger women.
https://www.redgifs.com/users/haaaardtimes

what AI did you use to make videos?

Kling.ai for almost all of them.

Can you please give some advice how to get comparable results? Your outcome is very, very good, but i would prefer to do it with smaller women

Beautifull i like

OMG redgifs is blocked in florida

I've been wondering how that Veo 3 would come out making good freeze and carbonite videos

I use HOTSPOTSHIELD. It's free and you can select New York or LA with less restrictions than FLA.

Videos created in Gemini

Partock wrote: Sat May 31, 2025 6:11 am

ensignronald wrote: Fri May 30, 2025 9:48 pm

neisy03 wrote: Fri May 30, 2025 9:41 pm

what AI did you use to make videos?

Kling.ai for almost all of them.

Can you please give some advice how to get comparable results? Your outcome is very, very good, but i would prefer to do it with smaller women

It's really a lot of trial and error. I have about as many less-enticing videos of dancing statues, imploding statues, and just other weird duds.

I've had the most success on Kling.ai, but that's probably because I've been using it longer. I used up my credits already with Google's Flow AI video tool, but I had a hard time getting the results I was looking for.

I always provide the initial frame because with each generation costing me money, it's best to remove as much mystery as you can. All images are originally made with Google's Image FX (and some more recently with Whisk, but I still don't have a feel for Google's new image AI in Whisk/Flow), sometimes touched up in Photoshop, then popped into the AI video site's initial frame slot.

Through some experimentation I've found some methods wrangling the video's actions. Having simply a frozen woman there and moving the camera around her gets easy once you just refer to the freeze subject as a sculpture or statue.

Loading up the Negative Prompts (Kling has it, I don't think Flow does) is also helpful, with my all-subjects-already-frozen videos I just get granular with it and say: "limb movement, facial movement, finger movement, body movement, neck movement, hair movement, mouth movement, eye movement".

Transformations, which I'd really like to do more of, is a lot more fraught. Even when going from two lined-up edited and made to be compatible images, both Kling.ai and Flow will have a hard or insane/uncanny transition. I've had like 2 successes and dozens of unnerving failures. Here's hoping that someone figures out how to make it work better.

I've also had some success with adding another non-frozen person to interact with the frozen subject. Like hug them, have them try to move the freezee, etc.

That's about it. It's a gamble, but you can get some nice stuff out of it.

ensignronald wrote: Sun Jun 01, 2025 10:10 pm

Transformations, which I'd really like to do more of, is a lot more fraught. Even when going from two lined-up edited and made to be compatible images, both Kling.ai and Flow will have a hard or insane/uncanny transition. I've had like 2 successes and dozens of unnerving failures. Here's hoping that someone figures out how to make it work better.

I've found Minimax (Hailuoai.video) to be surprisingly good with transformations.

https://hailuoai.video/share/ai-video/nAzWq2xQbpg0

It still takes a fair few attempts to get something decent, but its prompt adherence in general is quite good as well, so it's not as bad as some models which just seem to ignore the prompt entirely more often than not. They also do have a genuinely unlimited plan available if you've the money for it, so misses aren't costing you a limited supply of credits.

Unfortunately it's very limited in terms of other features compared to the likes of Kling or Flow; videos can only be six seconds long and there's no built-in method to extend them, there's no HD videos (720p is the max), and no advanced editing tools.

create videos and images ai https://pollo.ai?ref=mduxyju

my trials at WAN
cheers
yendis

yendis wrote: Sat Jun 07, 2025 8:23 pm my trials at WAN
cheers
yendis

wan video 25may25 [nadm 1].mp4_snapshot_00.03_[2025.06.08_09.18.12].jpg
wan video 25may25 [nadm 3a].mp4_snapshot_00.04_[2025.06.08_09.19.00].jpg
wan video 25may25 [nadm 4a].mp4_snapshot_00.04_[2025.06.08_09.19.26].jpg

Nice! I recognize where these are from. Care to share the videos? This could be promising.

You can create account at WAN --> https://wan.video/
checkin everyday to get free credits, relax mode can run but long wait

Links to my videos above:
video 1 --> https://drive.google.com/file/d/1NVh5iF ... sp=sharing
video 2 --> https://drive.google.com/file/d/1VHU_62 ... sp=sharing
video 3 --> https://drive.google.com/file/d/18ehOzM ... sp=sharing

cheers
yendis

Thanks for sharing!

yendis wrote: Sat Jun 07, 2025 9:30 pm You can create account at WAN --> https://wan.video/
checkin everyday to get free credits, relax mode can run but long wait

Links to my videos above:
video 1 --> https://drive.google.com/file/d/1NVh5iF ... sp=sharing
video 2 --> https://drive.google.com/file/d/1VHU_62 ... sp=sharing
video 3 --> https://drive.google.com/file/d/18ehOzM ... sp=sharing

cheers
yendis

how long does it take to generate images or videos? i tried free, and it is taking forever.

I tried the relax mode and it took 6 hours but it did create it in the end.

By the way, there's a more sophisticated AI Video service called Higgsfield AI.

https://higgsfield.ai/

They are really good at specialised shots and can even crossover two different types of shots in some instances. Plenty of examples before you commit to a generation. Current wait time for a 'Lite' (free) generation is around 1 hour.

Recommend you check out these ones: Bullet Time, 3D Rotation, Turning Metal, Medusa Gorgona, Arc Right, Arc Left, any of the Dolly or Crane ones. They generally good at keeping a character/subject still.

I'll upload any of my creations to this drive:

https://drive.google.com/drive/folders/ ... sp=sharing

https://drive.google.com/file/d/1ThstcZ ... p=drivesdk created for me en pixverse

yendis wrote: Sat Jun 07, 2025 8:23 pm my trials at WAN
cheers
yendis

wan video 25may25 [nadm 1].mp4_snapshot_00.03_[2025.06.08_09.18.12].jpg
wan video 25may25 [nadm 3a].mp4_snapshot_00.04_[2025.06.08_09.19.00].jpg
wan video 25may25 [nadm 4a].mp4_snapshot_00.04_[2025.06.08_09.19.26].jpg

YENDIS!! This is inspired by STuckfast/Guy Talon's "Never a doll moment". Its one of my favorite freeze films.
The poses are the same. you replaced them with asian women. I love it!

any free user of kling AI having trouble generating videos from images (or anything)?

it says that the server is busy, so free users temporarily cannot generate.