There's been many mentions on UMD over the years about wamming on a bed. While there seems to be many who enjoy doing and/or seeing it, there's usually comments about how difficult or impractical it can be.
For me, it can't be just a dull drab old looking bed in a poorly lit room, it has to be plump luxurious and ideally feminine like it's a woman's bed. Much like my love of women wamming and ripping/destroying business attire, it seems to be the juxtaposition of something prim and proper, elegant, in a situation it is even less likely to be in.
Of course, that adds cost to the equation. I've splurged a number of times in an attempt to create this, but ironically, I've had much more luck with non-WAM producers - but that's a story for another thread!

So isn't it great to have a rapidly developing tech like this to fill that massive gap!
Without further ado, here's some results of my experimentation so far.

I'm not sure if it's exactly matching the theme, however it reminds me of my dream something like luxury sleeping with elementals.

WAM in bed is one of my favorite scenarios. An outdoor luxurious bed would be a dream come true for me. Both to experience and to watch in a scene. There have been producers that used bouncy castles which seems like a similar scale in terms of effort for setup and cleanup, but beds seem more expensive because you cannot rent them. Maybe it is possible to create an outdoor bed scene with a green screen and indoor bed

I have similar dreams and fantasies about outdoor beds, in front of a lake or ocean, or under blooming trees. Both wam and non-wam.

How about on the beach, watching the sunset, during warm summer rain??!
Now that would be nice!!

It really is amazing how fast this technology is developing. It seems every few months we see another big step forward.
For me, Bing/Designer was the best balance of decent prompt interprestion, quality/realism and ease of access - but that damn dog! Others that had less fierce filters, either required elaborate local setups or just weren't that good in quality or realism.
Then FLUX came out...

That drew me back to CivitAI and with @messg's awesome Lora's, I was able to produce these "updated" images.
I'm still experimenting with prompt wording in Civitai's system - and I can't seem to get wet, rain, water depicted that well, help please! - but the early results are very promising!!!

They're really good. Is that using the custard_V1.1 Lora? I have to admit, the settings available on Civitai don't allow quite the same quality of mess as I the model produces locally but these are pretty good.

Thanks!
Actually, about half of them are the WAM Lora and half the Custard one.
The Gunge Lora hasn't been as consistent. I was actually getting better results using the Custard Lora with gunge prompts, than the Gunge Lora.
Did you include any wet/wetlook data when training the WAM Lora?

No wetlook in the dataset. The gunge lora needs reworking with the same parameters I ran the latest custard one with. I wasn't really too happy with it but haven't got round to retraining it. Same with the original WAM lora.
I'm testing an updated custard one which I will be uploading some time this week and currently training and updated pie lora. The gunge one will probably be next after that.

You are right in your observation though. The custard one should be be trained on the concept of getting messy as well as custard. I would expect other substances to be possible with it although not as good as a specific Lora.

One of my videos features wamming on a bed. It's only over a minute, so take a look:
https://umd.net/profile/i/bobographer/section/videos/video/francescas-pancake-bed

Nice! It's amazing to think that it's still early days for video ai.