Tantabus
Filters (Default) SettingsRegisterLogin
The creator tags have migrated. All "prompter" and "artist" tags have been migrated to "creator". Please check your watched tags, as they might have disappeared from your watch lists due to the change.
41 53 48 5 13
Subscribe
Galleries
Featured in
Sign in to create a gallery
Related
View VS Download DS
Uploaded by DocentHorsecock
 450x450 00:18.60 WEBM 3.92 MiB
Description

This is a cropped preview of next vid I am working on. Again this originally comes from Goblintide, 6 sec clip with Octavia. I will try to make 3-4 minutes out of it at 64 fps. This one contains only 3 segments, but you can’t really see that because I managed to remove the contrast/saturation/brightness blip that happens at the end of each segment. I generate 10-5s long segments and simply cut off last 0.5s and use the new last frame as the next input in next segment. It does reduce the resolution of input images, but it does not seem to impact quality that much. Not having these blips is great.
Compare this to Twi video, which has these blips.
Again, all the credit for the source vid goes to Goblintide.
I can generate like 30s a day (platform limitation), today I only got 20s of usable clips, others caused an error and terminated generations still consume time credits, so yeah. Probably will upload final version next week.
Enjoy her lovely expressions!

Tags
Edit History (4 changes, 24 tags)
explicit37719
creator:docent horsecock71
creator:goblintide298
edit870
editor:docent horsecock62
octavia melody900
earth pony15483
pony35085
g470943
ahegao2216
animated1225
bed9486
butt16954
cropped porn6
doggy style815
female77627
furless21
implied human on pony action61
implied interspecies79
implied sex151
implied straight450
looking at you32798
looking back9373
mare22422
moaning494
moaning in pleasure1026
night4609
no sound273
offscreen character9590
offscreen human1042
offscreen male8753
on bed5072
plot7917
tail16671
webm759
Source
Add
not provided yet
Report Share List favoriters

Comments

Edit Preview
Syntax quick reference: **bold** *italic* ||hide text|| `code` __underline__ ~~strike~~ ^sup^ ~sub~

Detailed syntax guide

13 comments posted
dirtyfinger

@DocentHorsecock
oof, really depends on the VA.
e.g. MagicalMysticVA takes $5 minimum and $0.20 per word, but everybody has their own payment schemes. Like Patreon with a request waiting list.
Just go to derpibooru, pick a video where you particularly like the VA, click on the VA artist tag and you’ll see their various web presences.
Or get started with these:
https://twitter.com/bakusatsuho
https://twitter.com/ChloeAngelVA
https://twitter.com/CottontailVA
https://twitter.com/DeltaVA_AD
https://twitter.com/EmpressBlueVA
https://twitter.com/EndymionVA
https://twitter.com/FlirtyFawn696
https://twitter.com/flufflabutt
https://twitter.com/JackieSpyceVA
https://twitter.com/KumBomb
https://twitter.com/LadyEliseVA
https://twitter.com/magicalmysticva
https://twitter.com/MidnightDatura
https://twitter.com/LewdWhorseVA (aka Wubcake)
The VA from melvelvin’s RD is Wubcake, btw.
Posted Report
Edited
dirtyfinger

@DocentHorsecock
that’s actually pretty cool. you are doing what i wanted to do but never got around to.
I’ll have to setup a comfyUI workflow eventually, cause I sure as hell won’t throw money at the generator sites.
Regarding the 15.ai sounds - it takes a bit of work and blending to make it convincing, but it’s quite doable.
here’s how my audio track looked like for an old vid of mine:
large
That didn’t use 15.ai but snibbles from a dozen different VAs taken from many clips - but the principle is the same.
Posted Report
Edited
DocentHorsecock

@dirtyfinger
Yes, I am generating new frames.
Compare the original Twi 6 second vid:

To my 180 second version:

I would not be playing with this if all I could do was just clever splicing of existing frames. Second version is worse from the generation standpoint because it’s not as jiggly and high quality, but it is longer and more varied. Still, all the credit goes to original prompter Goblintide, as I myself can’t generate anything like the original vid. I can only extract some frames, make a storyboard for 2 frame img2vid and connect clips to make a longer vid for fun.
I did try the oh-ah Octavia on 15.ai but results were kinda meh imo. Maybe I need to try again and remind myself how Octavia sounds like.
Posted Report
dirtyfinger

@DocentHorsecock
That is basically the idea, the aim is to make it more varied. 3-4 minutes is just a guideline. That Zipp animation I
made is like 12 minutes long, but it took like weeks to generate, as I have only 25 minutes daily on the GPU and that
yields maybe 1 minute of clips.
oh, you’re generating extra frames?
I usually just worked with what was there and used some video editing tricks to loop the things and even stretch them considerably if enough material is there.
If you can, you should do both. there are some simple video transitions that allow you to glue clips together without noticeable edit. like this one >>58147
For reference, making a non-looping clip loop and extending some scenes without noticeable cuts shouldn’t take longer than a couple of hours of editing.
I’m sure you can find plapping noises, too, but in case you don’t find any: Get a microphone and finger a tub of jello. 😀
@tyto4tme4l
It works better than I expected.
e.g. for Octavia, use the default settings and give her a script like “Oh. Ah.Uhm. Oof. Ohhhh. Harder! Ah. Ah. Ah! Aaagh!” and so on and pick the best ones. With generic female breathing sounds it can get quite convincing.
Posted Report
Edited
DocentHorsecock

@tyto4tme4l
While I was aware about 15.ai I did not know that it can do good moans.
@dirtyfinger
Oh wow, not bad sources. I might play with this one day. Ripping audio myself is out, as I don’t really want to invest reaserch time into that. But using existing audio could fit within the time I can invest in this.
I will look into these things, thank you both! One more thing I might need are plap sounds, kek.
@dirtyfinger
Not sure why you would want to stretch it to three minutes … resequence to as much variety as you can, then just upload a seamless loop.
That is basically the idea, the aim is to make it more varied. 3-4 minutes is just a guideline. That Zipp animation I made is like 12 minutes long, but it took like weeks to generate, as I have only 25 minutes daily on the GPU and that yields maybe 1 minute of clips. With cowgirl and missionary it was easier, because doggy style tends to cause neck breaks when the character turns the head way too much, so some clips get rejected.
Thanks for help!
Posted Report
dirtyfinger

@DocentHorsecock
man, brightness and color glitches are so hard to correct. but this clip barely has any strong changes so you should be fine.
Not sure why you would want to stretch it to three minutes … resequence to as much variety as you can, then just upload a seamless loop.
For sounds you could rip some out of the show itself. there’s at least a couple of usable groans in episode 100 that would fit. for variety, just rip sounds from other existing clips from derpy and adjust pitch and speed.
I wish I still had my little collection of naughty sounds, but a drive crash killed all my old projects.
For clean samples without hassle, you could consider getting a few moan packs from various VAs.
https://magicalmysticva.bandcamp.com/track/rarity-moan (rarity moans make a good octavia substitute)
https://www.101soundboards.com/users/lgqzb-magicalmysticva
Pro tip: don’t neglect breath sounds. they are nearly universally usable and add a lot.
And yeah, 15.ai has gotten surprisingly good with ‘Oh. Ah. Uhm. Oof. Ohhhh. Harder!’ etc. sounds.
Posted Report
Edited
Background Pony #D466
I assume we will not get any audio in the final version, sucks because my fascination with Octavia has a lot to due with her English voice. And, well, yes, my pony fetish.
Posted Report