Today, I’m here to introduce a free website that can create dynamic videos from text/imagesRunwayRunway can generate images from text, transform existing images, remove backgrounds from videos, and eliminate people or objects from footage.
“Runway”, a free website that generates dynamic videos from text/images
Runway has many features, including background removal, text-to-image, image-to-image, image expansion, and removing people or backgrounds from videos. Today I’m going to introduce a few of these to you. If you’re interested, feel free to click!hereTry it yourself.
Text-to-Video
I use prompts I found online to have Runway generate videos for me. For example, when I input “War mecha,Full-body photo,girl standing in front of a large robot, in the style of epic fantasy scenes, Face shot,32k uhd, serge marshennikov, realist detail, adonna khare, emek golan”, Runway generates a 3-second video. The girl in the video can move around, and smoke comes out. But! The girl’s face only lasts for one second AAAA∑(ι´Дン)ノ, after one second, once this girl moves, she becomes a faceless woman. You can copy this prompt and see if your video has this problem too.
Besides the video of the robot and the woman that was just shared, the editor also tried another oneRegarding the image of a woman and a cat, I entered the prompt “a hair women and a Giant cat are in the car, orange cats, the cat has Big brown eyes, krenz cushart, unreal engine 5, hikari shinoda, realistic detail, radiant clusters, pseudo-infrared, minimalist, 16k, best quality”, Runway did a pretty good job – both the woman and cat were rendered in the car, but no matter how I look at it, it seems like the woman has ears on her head. I don’t know if it’s just my imagination or not – do you all feel the same way?
Image to Video
Besides text-to-image, Runway can also do image-to-video. The editor here uses my cat, Salang, as a demo. You can see a cute orange cat, and the video features changing light angles, with Salang’s face even turning to the side! Though it’s not very natural, it’s still impressive and deserves praise. However, this side profile doesn’t really look like Salang because Salang doesn’t have such a wise gaze.
Of course, there will be failures too. Sometimes the results from the same photo just won’t turn out as expected. Like in this video, there’s no sense of movement, just simple light and shadow shifts and just a tiny bit of a side profile. So if anyone makes something they don’t like, feel free to try a few more times!
Generate videos from text and images
Editor input andThisThe same prompt “a hair women and a Giant cat are in the car, orange cats, the cat has Big brown eyes, krenz cushart, unreal engine 5, hikari shinoda, realistic detail, radiant clusters, pseudo-infrared, minimalist, 16k, best quality” was used with photos of both Salang and myself (heavily retouched but posted anyway since I paid for it), yet it still couldn’t generate a video matching this prompt. Here’s the video created by Runway using this prompt with Salang’s and my photos. The Salang version only gives off the impression of light streaming in, while mine doesn’t seem to have changed at all—completely lacking any sense of a cat or a person.
The editor refused to give up, changed the Prompt to “a hair men and a Giant cat are in the car, orange cats, the cat has Big brown eyes, krenz cushart, unreal engine 5, hikari shinoda, realistic detail, radiant clusters, pseudo-infrared, minimalist, 16k, best quality,” and paired it with photos of Dad and Sa Lang. As a result, Dad’s face went completely haywire, while Sa Lang did actually get transformed. Poor Dad though.
Although Runway has some issues with text-to-image and image-to-image generation, it still stands out in video processing. Previously, the Computer King Ada YouTube channel already posted a tutorial on the free AI video creation tool Runway.filmIt covers details on video background removal, object removal, and video upscaling. Check out the video below to learn more if you’re interested.
Source: KOCPC Chinese