AI Video in 2026

Published: Apr 7, 2026 by Isaac Johnson

A user recently requested by way of the form at the top that I look into Kubrix AI. I wasn’t sure if it the person was involved with the project or genuine interest. Either way, I warned them I would do an honest take.

While I’m at it, I’ll try a bunch of the prior AI video generation sites I’ve used (Kling, Pollo) as well as my tried and true Gemini and Midjourney. I’ll also see what I can find that is new (and most important “free”) while I’m at it.

Kubrix

First, the splash screen makes me want to vomit.

I gave it a prompt and then picked a size and duration

/content/images/2026/04/kubrix-02.png

I used Analog File for a style

/content/images/2026/04/kubrix-03.png

lighting style

/content/images/2026/04/kubrix-04.png

A frame

/content/images/2026/04/kubrix-05.png

Then a model

/content/images/2026/04/kubrix-06.png

and an optional start image

/content/images/2026/04/kubrix-07.png

Lastly, a Generate page that says this 4-second video will be “1800 credits”

/content/images/2026/04/kubrix-08.png

It showed it would use the prompt:

A cinematic 4-second opening sequence for a modern ALF reboot. The shot begins with a rapid, smooth dolly-in across a sleek, contemporary suburban living room bathed in the amber glow of a California sunset and accented by cool teal and magenta smart-lighting. ALF, the iconic alien from Melmac, is perched on a minimalist charcoal velvet sofa, looking more realistic than ever with high-fidelity, hand-groomed brown fur textures and expressive, glossy eyes that reflect the room’s ambient light. As the camera zooms in, ALF nonchalantly tosses a high-tech tablet aside, turns toward the lens, and breaks the fourth wall with a signature mischievous smirk and a quick, knowing wink. The atmosphere is vibrant and polished, featuring professional 8K resolution, shallow depth of field that blurs the modern kitchen in the background, and crisp, cinematic color grading. In the final second, a sleek, chrome-textured “ALF” logo with a subtle digital glitch effect emerges in the foreground, pulsing with retro-wave energy.

I signed in next and see as a new user I get 100 credits and the new price is now only 1000 credits

/content/images/2026/04/kubrix-09.png

This leads me to “how to get credits”.

I can either subscribe to a monthly fee, the lowest of which is US$30

/content/images/2026/04/kubrix-10.png

or a one-time top up for $40

/content/images/2026/04/kubrix-11.png

I was willing to deal with at most US$10 to try a thing so unfortunately we will have to pass.

Midjourney

I was curious what Midjourney would do with that prompt. Even though I mostly use Gemini now when I need a bit of AI graphics, I still pay for MJ as it was the first that really offered solid AI Generated art.

As you can see, it’s generally best to get some images then run a video from the best of them (if there is a best of them)

I tried the same prompt, but this time fed it an ALF still to start

The text is a mess, but I was surprised how it accurately figured out how Shumway talked just from a still

and a regen

Gemini (VEO2) video

Let’s put the same thing in Gemini to see what it comes up with

Here is the video (with some strange sound effects)

I gave it an image to refine the video

/content/images/2026/04/gemini-03.png

which did better visually

One last refinement. I asked “Can you give him a deeper New York snarky comedic voice”?

I tried again the next day…

When I asked it to tweak the voice for a NJ accent, it re-imagined the look again

This highlights one small limit. Using Google AI Pro (which I got a year of with my last Google Pixel Fold phone), I get 3 videos a day. So to do more, I either pay out of pocket in AI Studio in GCP or I just wait till tomorrow.

Kling

Next I gave Kling a try

/content/images/2026/04/kling-01.png

But even though I have sufficient credits, it would seem they paused the free tier

/content/images/2026/04/kling-02.png

I came back a couple days in a row just to check, but it was still blocking free tier.

Self (ComfyUI)

As we’ve shown, using something like ComfyUI on a computer with a GPU, one can just create these kind of things locally.

I fired up ComfyUI.

I’m going to use a Wan Vace 14b template I have bookmarked.

It might looking daunting, but really I just need to give a seed image and a prompt (see the arrows)

/content/images/2026/04/comfyui-01.png

I’ve translated those Chinese statements in the negative block before and they are pretty canned “no watermark, no junk, etc…”.

I then tried with LTX2 (needed a 40Gb model downloaded)

/content/images/2026/04/comfyui-02.png

a little weird but could be a start

I tried LTX 2.3 which can take a video input with the same prompt and ALF still

/content/images/2026/04/comfyui-03.png

Though it gave me the oddest output

A second pass with just asking for a “The TV intro for an ALF sitcom reboot.” (but also with the still)

One last go, I set the prompt to “The TV intro for an ALF sitcom reboot. Should include this character”

Seevideo.ai

I found an AI Video site, seevideo.ai that would give me just enough first-time credits (100) to try Seedance 1.5. It isn’t the latest 2.0, but it was worth a try

/content/images/2026/04/seevideoai-01.png

Dreamina capcut

Another one that gave me just enough for Seedance 1.5 Pro was Dreamina

/content/images/2026/04/dreamina-01.png

Pollo AI

Pollo.ai has it’s own model as well as others. I’ve used them in the past for some nice intro videos.

/content/images/2026/04/pollo-01.png

It was interesting how it showed the still but then just did it’s own thing

Seedance2ai

One that wouldn’t let me use Seedance2 (for pro customers), but did let me try the now defunct Sora2 was Seedance2ai

/content/images/2026/04/seedance2ai-01.png

But I have a feeling it will never complete

/content/images/2026/04/seedance2ai-02.png

In the end my generated workspace disappeared (but didn’t charge the credits)

Summary

We tried many many AI video generators today. It is not that surprising that we can find an endless assortment of new AI startups that will give just enough credits to try a video or two so I question those video generation sites that give new users nothing. There is no real motivation.

This post started with a user request to review Kubrix AI. Since they offer me nothing and have this awful background video, I can say confidently I won’t be returning. Of the new ones, the closest I got to paying was Seedance2ai.io as if I dropped the coin for an annual plan, I could have used Seedance2. That said, soro2.ai would let me if I dropped a one-time $50 top up.

I could have used PowerDirector as I have 100 credits in there as an annual subscriber. but they didn’t offer any unique models I wanted to try

/content/images/2026/04/powerdirector-01.png

If I was tasked on making a TV intro with just the tools at hand, I would likely splice some MJ clips together and use Suno to make a Jingle…

/content/images/2026/04/suno-01.png

then create with a style

/content/images/2026/04/suno-02.png

Something like

GenAI kling pollo seedance gemini ComfyUI midjourney

Have something to add? Feedback? You can use the feedback form

Isaac Johnson

Isaac Johnson

Cloud Solutions Architect

Isaac is a CSA and DevOps engineer who focuses on cloud migrations and devops processes. He also is a dad to three wonderful daughters (hence the references to Princess King sprinkled throughout the blog).

Theme built by C.S. Rhymes