#viducpp

[deepdreamnights]

AI Video Development Demonstrated Via Music Video

Okay, so, I thought it would be a good idea to show how AI video has developed over the past two years by comparing the music videos I've made with it over the same period.

July, 2024 - Pop-Pop-Bagok!

First real attempt at an AI music video. Clips at the time were short, limited in motion, and low-res, so most of this is stills and editing.

October 1, 2024 - Are You Prepared?

Another that's mostly made with clips from civil defense films and other archive.org scroungings, but clips were getting easier to acquire for free, and could get wilder in the end results. First time I was able to do lip-synch.

October 30, 2024 - Parasaurolophus

As you can likely tell, stuff jumped up quite a bit between October 1st and October 30th, as Parasaurolophus is done entirely with AI base footage. This one was largely if not exclusively an img-to-video project, allowing for the weird creature designs and environmental prettiness.

Clips maxed out around 4-5 seconds at this stage. Anything longer had to be screen-capped from the last frame, then started from that in a new gen and sewn together manually.

It's around here that I got into the Vidu artists' program, and most of what's done here and beyond is with Vidu primarily.

Nov 29, 2024 - (Lets Dance to) Dinowave

We start getting ref-to-video with this one, clips going longer, and strong improvements with coherence. Around this time the systems started being able to handle dinosaurs more consistently, so of course I leapt on it.

Improvements are obvious, what might not be is how much I'm cutting around glitches and errors.

Jan 27, 2025 - An Open Letter to the SCP Foundation

Stock footage is still needed at this stage, as I used it for the typing sections. Reference-to-Video takes a much higher load of the process here, and Director Kleinheart is the first character to be a central focus of the video because consistency reached the point where that was viable.

Still a lot of quick cuts and edits, posters sliding over to hide awkward transitions, etc.

Feb 15 - Rom & Brandy/Ballad of the Spaceknights

A very similar situation to the previous, but with a lot more compositing and post-production work, as on its own the AI leaned too modern in the special effects department so I did them largely myself. For several shots, I generated characters on a greenscreen and composited, a technique that becomes increasingly useful for complex scenes down the line.

April 29, 2025 - Synth Night in 87

still doing manually linked img-to-video to create long shots, this is my first unbroken shot video. Every single one had to be speed adjusted manually as momentum couldn't carry over.

May 10, 2025 - It's Bismuth, Baby

Using the same endless shot technique from Synth Night, for this one I generated each shot, edited it in, and then moved to the next shot, which combined with editing and speed manipulation let me time things out to the song and reproduce something resembling my in-my-head music visualizations in video. A technique I'd use later with "Space Seals"

I timed the clips to end on cymbal clashes as much as possible so I could hide transitions under the visual "jolt" timed with the cymbals.

Skipping to-

October 28, 2025 - Lady of the Maze

This video was my inaugural test of Vidu's character-reference feature in a music video (actual first major use was in After-Cool Action Blok but that's not a music video). This is also the first chance I got to really take advantage of extending shots.

Feb 9, 2026 - If the Goo's Alight, it's Not Alright

Both this and Ain't No Basic Beast were made with the same techniques under the same new Q2-pro & Q3 models. But If the Goo's Alight is more ambitious and has a lot of dialog whereas ANBB has none, so its my use of the 10th video Tumblr allows me.

It's also a good conceptual bookend with Pop-Pop-Bagock, as both are weird sketch ideas that were previously outside my scope and both are meant to be corporate artifacts from a bygone age.

A smidge under 2 years of video model advancement, through the lens of extensive editing and cutting.

I wish I had PPBG's editing file handy, but for reference, this is what the Goo's Alight editing timeline looks like.

So in case it wasn't clear, one thing that remains a myth is the 'push button, get final result' one.

[melinoelabs]

T.O.B.O.R. needs your help-

-to help you help yourself.

Or at least this missive would indicate that. If you are a human being trapped in or around Site 741-45, and you have not seen Venus, maybe you can help.

Lyrics, design, and editing by S.T.T. Music: The DeepDreamNights Robot Band, Via Suno Visuals made with Vidu q3 ref-2-vid, try it here with code VIDUQ3R2VCPP for 100 free credits.

Technically sponsored content as I'm in the Vidu Creative Partner Program.

[deepdreamnights]

In 1989 teen monsterism was at an all time high.

A situation that was made substantially worse by the release of:

If the Goo's Alight, It's Not Alright!

Which introduced (advertently) the world to Goozy, the anti-goo mutant monster and (inadvertently) to the idea of drinking mysterious glowing liquids, which had not occurred to many people prior to the campaign.

Ultimately, four American cities were laid waste by the resulting "muteens", making it the most successful/least lethal PSA campaign of the 80s and 90s.

(Made, with Vidu's Q3 & Q2-Pro models, try it out free here + get 100 bonus credits, and I am in the Vidu Creative Partner Program.)

[deepdreamnights]

A little something for all you 00s kids who grew up shipping Audi and Hellhop.

-

Actually, this is the first of several videos I made to help showcase what Vidu's new Q2-Pro (this vid) and Q3 (upcoming ones) can do.

The short version?

I don't have an excuse not to make my shows anymore.

And chances are, you might find yourself in that same boat. Sign up for free with my code (ROBOTQ3) and get 100 bonus credits and give it a shot.

While I've had the Daem-On characters in my back pocket for awhile, and the song as well, the actual above video was about 36 hours from start to finish, with the usual manual video editing and manipulation mixed with Vidu's new ref-to-video model.

The big improvement with Q2-Pro is the ability to use videos of up to 8 seconds as reference prompts, which can be used to tighten up/fix problems with otherwise good gens, update vids made with older gens, and can be used as a reference for character and camera movement.

Beyond that, Q2-Pro gives a lot more coherence, as several of the Daemes simply would not work properly under previous models.

Detailed tutorials are, of course, to come.

[deepdreamnights]

TyrannoMax & The Warriors of the Core Sizzle Reel

Nothing is final, obviously, it being a sizzle reel.

But you get at least brief introductions to TyrannoMax, BrotoSarah, DeinoSteve, TriceraBruce and PteroDarla, as well as a few choice moments with some of the more fun villains.

Who's your favorite?

Worked up with Vidu Q3 (try it for free, etc) and a ton of built up TMax assets. My YT channel is here.

I'm in the Vidu artists' program, and all that suchlike, consider it spon-con.

[deepdreamnights]

(un)Real(ality) Cryptid Encounters

Another Vidu test, this time primarily working with the text-to-video model, with a few image-to-vid sections, with some basic video editing.

The sound effects are native to the gens.

[deepdreamnights]

5 Years ago, temporal refugees from the Cretaceous period began arriving. Now, we catch up with them to see how our newly rediscovered cousins have adapted.

--

You may recognize these dinos from here:

Which also lets you see how far things w' Vidu have come in a year.

Made with Vidu-Q3, image-to-video model, I'm in their artist's program.

Details and Tutorial under the fold.

Looks like I don't have an excuse not to make those shows now.

Vidu just launched two models, Q2-Pro with video-prompting and reference-to-video, and Q3, which is currently text-to-video and image-to-video only, but can do up to 16 seconds with dialog and sound effects.

As the video above is longer than 16 seconds, obviously, editing is involved, as it always is when something is more than a quickie one-off image or clip.

But, this does open up a lot of potential.

The basic process of putting together a clip is simple. You start with an image, and then you script out your prompt.

Long time fans might recognize these dinos and some of their interview quotes from here. So I had character art ready, and my own writing to start with.

Generally, where I started actually wasn't in Q3, but Q2-Pro, where I'd use the reference-to-video options to create starting frames. This is helpful because even with reference the AI has to reinterpret things and there's always some style shift. Using the ref-to-video to create base frames makes it so scene-to-scene those changes aren't random, as you're essentially giving the AI an image that's ready-made for Vidu's use.

Vidu even gives you a built-in method to send the images directly to the prompt with its frame extraction feature, but I recommend to always also download it (without the watermark), because you will want to go back to that image, have the option to edit it, etc.

From there, it's a matter of prompting:

Dynamic anime fight scene, multiple cuts and angles. 2005, traditional hand-drawn cel animation, masterpiece sakuga cinematic animation quality. A green-skinned, skullfaced barbarian queen with a huge scythe duels against an orange-skinned goat-anthro with pink flames for hair and trim fighitng with a golden trident. the goat-anthro wears a sparkling burlesque costume and has a long snake-like tail. SHOT 1: the camera circles the fighters slowly, establishing their relative positions and forms. SHOT 2: cut to closeup on the skullfaced barbarian, her voice is sinister and high pitched, dripping with sarcastic evil SKULLFACED: "A goat that's self-braising, how very quaint!" she is voiced by Sarah Skeletoria. SHOT 3: cut to the goat-anthro, front view, middle-closeup, she takes a defensive stance with her trident. GOAT-ANTHRO: "I won't let you hurt anymore dames, Skullsa Doom!" she has a valleygirl accent and sexy, confident voice. SHOT 4: Goat-anthro charges Skullfaced with her trident forward, Skullfaced barely dodges, the trident rips skullfaced's cloak. flame-themed anime motion background. She is driven by righteous fury. GOAT: "And I won't let you hurt Audi!" SHOT 5: Skullfaced swings her scythe, but goat-anthro backflips away, dodging just in time. her flame-hair forms a long arc as she dodges, enhancing the sense of grace an motion. SKULLFACED: "I don't require your permission, Baph-lessk!" SHOT 6: They continue to duel, evenly matched.

In the form of a script, essentially. You'll notice I used the same reference names for the characters, and that I avoided pronouns. This helps reduce confusion. As does phonetically spelling out stuff that's difficult to pronounce.

There is an important early-series bug that needs to be advised on. If the clip isn't the right length for the required dialog, the characters will often flub or babble lines. This means that you're going to want to time-talk-out your dialog before generating (Q3 is credits-hungry compared to Q2-Pro), or you get what I got here:

Which is the clip going off the rails partway through. If you're trial-and-erroring it will get expensive, but with editing, and by re-dubbing with recorded voices, you'd get around that easily.

Generally, I just generate shorter clips for editing, but for showing off the system, a full 16 second demo seemed best.

Vidu does improve their models over time, and I expect this to be ironed out in the next few weeks, but in the meantime, keep it in mind.

The potential of the tool is immense. Not just because of the coherence and quality of the animation produced, but because it's set up for actual production. A team of a handful of people could produce an entire show on a scrap of a budget with this.

If you want to try it yourself, go to Vidu here and use promo code ROBOTQ3 for 100 free credits.

This post is, I'm pretty sure, easily under the umbrella of paid promotion.

[deepdreamnights]

Synth Night in '87 (in the Haunted Vaporwave Woods)

A music video made (mostly) using Vidu's new ViduQ1 model and one of my Suno tracks.

Details and process under the fold

As a member of Vidu's creative partner program, I was able to participate in the beta and got a lot made during that window, so there will be a lot more about Vidu's new model soon.

Currently, the new Q1 model is a 1080p, 24-frames-per-second, 5 second model that functions on the image-to-video and text-to-video processes, with reference to video hopefully to come sooner than later.

I took an assortment of vaporwave landscapes I'd made in midjourney and used them as start-and-end point images, along with prompts along the following lines:

walk through POV footage, live action film of a surreal forest filled with strange glowing plants and fungus

With img-to-video, minimal prompting works fine for something like this.

The quality is a big jump from version 2, with a lot less blurring and smudging, especially with animation. For a lot of actions setting the movement amplification to large is needed, however, since Q1 is conservative with its movements.

In terms of quirks, start-and-end-frame image-to-video tends to produce a less crisp and less controlled result than just using a start frame.

As there is presently no "extend" function on vidu videos, the motion speed and carryover changes clip-to-clip, so doing a long shot like this one requires adjusting clip speeds, chopping the start and end frames, and the like. Even so how details are interpolated can change shot to shot, requiring a lot of tinkering to minimize the clip-transitions.

I added some overlays to assist with that, and the entire sequence is faster than the original footage in general. The dragon's fire-breath, too, is post-production compositing.