How am I making these videos?

[Condor621]

I've had several people say, "I don't know how you're doing this...", so, let me give you all a quick rundown. There are MANY people here who have been creators for a long time, with skills in Photoshop, Adobe, and all sorts of programs that I can't even begin to figure out right now... I WANT people to be able to make new content and videos, particularly people who are far more skilled than me, I'd love to see more of what WE want instead of the usual crap like "WIfe Like" and other nonsense we get.

SO, here are the tools I use to make these videos, and trust me, I started without a clue, cutting and pasting pictures in a photo editor, not understanding how to get the right tone of voice, inflection, and how to get consistent character appearances - that's actually been the toughest part until recently. Here's the stuff I use currently (I receive no compensation, am not affiliated with, or have any reward from these apps/sites, I'm just a hobbyist making nonsense..lol):

FOTOR is a photo editor, with a TON of handy and intuitive Photo Tools like background removers, ability to tweak things, add elements, text to image AI capability, and even swap faces - this was a HUGE addition to get the characters to look as consistent as possible from scene to scene. Abigail was the first story I made where she pretty much looked the same from scene to scene.

app.artflow.ai is AMAZING. You can upload images of a person or AI generated image to create an "Actor", and then you can make unlimited characters with the same face, body, etc. This is a game changer, and I only started using it in the last week or two. Like everything else, it has a cost, its $9 to create an "Actor", but once you do, you can have the same face and body as a Fighter Pilot, cop, fitness instructor, whatever. The next 2 parts of my "Nikita" story will be using this feature, and its been awesome to work with. Its really fun to work with and to tweak your results. The "Add character/Actor" to the text to image tool is great.

ElevenLabs: This is where I import voices from into the next tool. Most Online AI voices are, well, AI voices. No inflection, incorrect inflection and emphasis, and no way to control anything beyond pitch and speed. ElevenLabs Multi-Lingual voices are great, Stability, Similarity, Style, and speed allow so much control. Every video I put together, I listen to each scene and adjust the sliders to get the right tone and style I want from the characters. Its tedious, but again, I like doing it.

HeyGen : THIS is the tool that brings it all together. HeyGen is a Photo to Video creation tool. Its also the most expensive part of the process, but without it, I'd be back to Pictures and audio files on SoundCloud. HeyGen allows you to upload video or photo assets, and upload photos to animate. Basically I figure out a basis of a story, try to create the right characters, and then upload the photos to HeyGen. Then, its just Text input and the app synchronizes and animates the photos or images. There are some tricks to it, figuring out how to get multiple characters in a picture is tough because each scene only allows one speaking part. I imported my preferred voices from ElevenLabs, tweak them, and write and edit until it l sounds good, then I submit for Animation.

CapCut: A free, Online video editor. I download my completed HeyGen videos (after watching and ensuring they came out OK) and upload them to CapCut. Here, I can add Video elements, overlays, effects, sound FX/Music, etc. from Downloaded videos or available stock assets. Then, when its clipped and edited, I export, Download, and upload to Streamable for sharing.

Its work, can be a bit tedious, but its a fun hobby. Yeah, there are costs to some of the sites (HeyGen works on a credit system, and has a couple of levels to work with) and it can get pricey, but I guess its better than a lot of other stuff I could be blowing my discretionary savings on.

Well, thats it. If you DON'T like AI or creators using AI< then I just ask you to hop on out, I do not want, need, or will listen to people lecture me about AI, nor will I stop. There is plenty of content here and on the WIki that I do NOT like... I just hit the back button and keep my criticism or snark to myself, I ask you to give me the same courtesy.

Good Luck and Happy New Year!

[Windblade]

How much time (approximately) it takes to create one video?

[Condor621]

Geez, it really depends. Some of them a full day... some of them, a few days for a 10 minute vid. I'm trying to make them better and better progressively and take more time with some. I'm working on another project, and STILL working on it for about, oh, 3 months now, the full continuation of the Hana Story, its a full length Military Sci Fi Movie, and getting it where I want is taking serious time. But knocking out a single scene (Bad Robotgirl, for example) is a couple of hours. But once I get a kernel of a story, it just kind of takes off from there. I research computer terminology to make things as accurate and "realistic" as possible, for the one with the 1950's Robot, I actually researched Univac and period specific computer technology and tried to make the visible components period accurate. I know, insane.

Just FYI, the next couple of installations of the Nikita Story are my best EVER. I love the character, like the world, and can honestly make a LOT of episodes in that story. Thats the other weird thing, I outline and write out the characters and stories beforehand, and if I really like a character, I want to make more vids of her. I really liked Abigail, she was cute and sweet and had spirit, and I was going to take it further, but I just didn't get enough positive feedback to keep up the enthusiasm, so I threw together the one video of the Roboslut as a quick "give the people what they want" thing, and then Nikita just... happened.

I can make Nikita into a damn ongoing series in the world I'm building, think of a beautiful Robot Companion model that's been illegally upgraded in a "John Wick" type near-future universe of space travel, criminal syndicates, and androids... James really pissed off some bad people.... LOL.

[smalk]

Thank you for the explanation, can't wait to see the new videos!

[D.Olivaw]

Thanks for the rundown, Condor.

For vocal stutters and glitches, I highly recommend doing the ElevenLabs script "clean" and then adding them in via your preferred sound-editing tool. The AI TtV tries to render them as normal speech, which both isn't really what we're going for when we do this kind of thing and (I suspect) messes up the rest of the voice read because of the way it affects the interpretation of the overall text.

If you want some pointers/technique help, let me know. I messed around with it for quite a while before and, maybe more importantly, after I did the Home Trial audio story and learned a lot the hard way.

Somewhat unrelated, but have you noticed a quality drop in ElevenLabs outputs? No matter how I set the sliders it seems to lose track of tone or mispronounce words far, far more often than when I did the clips for HT, to the point that I have largely stopped using it.

[Condor621]

Thanks for the rundown, Condor.

For vocal stutters and glitches, I highly recommend doing the ElevenLabs script "clean" and then adding them in via your preferred sound-editing tool. The AI TtV tries to render them as normal speech, which both isn't really what we're going for when we do this kind of thing and (I suspect) messes up the rest of the voice read because of the way it affects the interpretation of the overall text.

If you want some pointers/technique help, let me know. I messed around with it for quite a while before and, maybe more importantly, after I did the Home Trial audio story and learned a lot the hard way.

Somewhat unrelated, but have you noticed a quality drop in ElevenLabs outputs? No matter how I set the sliders it seems to lose track of tone or mispronounce words far, far more often than when I did the clips for HT, to the point that I have largely stopped using it.

Dude, if you could let me know how to do the "stammering glitch" speech, that would be AWESOME. I just use repetitive words or partial words and then use punctuation to get as close to what I want as possible. With the cross platform nature of what I'm doing (HeyGen, EL, CapCut, etc) I think it would be super difficult to edit speech that stays somewhat synched with the videos. But if theres another editor with better tools than CapCut, let me know. I know a lot of people use Adobe, but when I pop that open and try to look, I literally get lost. No idea how/where/what tools to use, so I've stuck with what I know.

Thanks and thanks for the support!

[D.Olivaw]

I use Audacity for sound editing. It's free and relatively simple to use since it's solely for that. Stammers and repetitions are quite easy to do, though there are some wrinkles to get them to sound good. You just look at the waveform to find the bit you want to snip and then repeat it. You can similarly stretch or compress things out, glide the frequency up or down, and add effects tracks. I might be able to find time to do this for you a couple times as an example if you shoot me some voice reads.

You're right that sync is the critical thing. One option is to run through the whole process clean, edit the audio track, and then either implement the same stutter (or, simpler, freeze the video) during the relevant portions in a video-editing tool. Another option would be to have two scripts, one clean one you give to ElevenLabs, and another that you modify after editing the audio to match the new, glitchy version. For instance, if the character stutters a syllable three times, you'd add that stutter to the script. I recommend going this direction, from clean to edited, because there's some "feeling out" of what sounds good when you do the sound editing that you won't be able to pre-plan in an initial script. Whether this second technique is viable depends on whether the tool you're using to animate the video can get the sync right with that stuff in the script.

[33cl33]

Somewhat unrelated, but have you noticed a quality drop in ElevenLabs outputs? No matter how I set the sliders it seems to lose track of tone or mispronounce words far, far more often than when I did the clips for HT, to the point that I have largely stopped using it.

I was getting some really bad results for a while... it seems to have more to do with the particular voice chosen. Though it burns through credits, I spent a ton of time testing and generating different ones to find more expressive / stable voices.

It also helps to generate longer sections at once. The more content it has to chew through at one time, the more narrative context it has to play with.

[Windblade]

Can I ask some technical questions (about video file size, video resolution, etc.)? I'm sorry if such questions are rude.

[Condor621]

I use Audacity for sound editing. It's free and relatively simple to use since it's solely for that. Stammers and repetitions are quite easy to do, though there are some wrinkles to get them to sound good. You just look at the waveform to find the bit you want to snip and then repeat it. You can similarly stretch or compress things out, glide the frequency up or down, and add effects tracks. I might be able to find time to do this for you a couple times as an example if you shoot me some voice reads.

You're right that sync is the critical thing. One option is to run through the whole process clean, edit the audio track, and then either implement the same stutter (or, simpler, freeze the video) during the relevant portions in a video-editing tool. Another option would be to have two scripts, one clean one you give to ElevenLabs, and another that you modify after editing the audio to match the new, glitchy version. For instance, if the character stutters a syllable three times, you'd add that stutter to the script. I recommend going this direction, from clean to edited, because there's some "feeling out" of what sounds good when you do the sound editing that you won't be able to pre-plan in an initial script. Whether this second technique is viable depends on whether the tool you're using to animate the video can get the sync right with that stuff in the script.

Thats is so helpful, and like a lot of what I'm doing, when I figure something out finally I say, "FFS, why didn't I think of that 3 months ago!" LOL.

Thanks! If you want, I wouldn't mind setting a Collaboration at some point. That might be pretty damn cool. I'm more of a "Spy Thriller/Military writer" with Sci fi thrown in, but your writing is far better than mine... in MANY respects.

[galactus_jones]

I am trying out HeyGen, but there doesn't seem to be any way to reload credits except by buying a new plan?

Or can you sign up for the monthly plan, then if you use up a month's credits cancel and start/par for a new plan?

[33cl33]

The only way to get more credits is by upgrading to a higher tier. The creator plan can be scaled in 15 / 30 credit increments up to 90 credits/month for $180: https://www.heygen.com/pricing#compare

Maybe with a VPN, you could try duplicate accounts?

The credits are also use 'em or lose 'em each month, and cost more in the more expensive business tier. Kind of a terrible pricing scheme, given how few minutes you get per month. I can't imagine what it would cost to use this for even the most basic enterprise solutions they propose it for (like personalized marketing emails).

[Condor621]

The only way to get more credits is by upgrading to a higher tier. The creator plan can be scaled in 15 / 30 credit increments up to 90 credits/month for $180: https://www.heygen.com/pricing#compare

Maybe with a VPN, you could try duplicate accounts?

The credits are also use 'em or lose 'em each month, and cost more in the more expensive business tier. Kind of a terrible pricing scheme, given how few minutes you get per month. I can't imagine what it would cost to use this for even the most basic enterprise solutions they propose it for (like personalized marketing emails).

I have a lot of issues with how HeyGen is priced and their plan options, and some of their other features. I particularly dislike that you can't see the finished Animations until you submit and burn credits, and then if you need to edit it, its considered a totally new submittal. But for the content I'm trying to do I haven't found a better option. As I said, these Vids are NOT cheap to make... But as tools refine and this type of thing becomes more widespread in use, I think it will get better... and hopefully cheaper. We shall see.

[galactus_jones]

So, you limit you production to 30/minutes (or whatever) a month? The pricing scheme is insane. I have started making some ASFR videos on the site, but I see that I could hit that monthly limit pretty easily. It would be nice if I could just fork over money for more credits (like Airflow- thanks for the tip)

Also, I was trying to do some more hard-R rated content and noticed that the content filters are pretty extreme on heygen. I understand they don't want people making videos of Taylor Swift spouting filth, but like you I don't really see a legit business use for this thing.