Am try to using Veo 2 to generate the video by image. Model is really cool i think, but the quota and safety setting (i know is necessary) but is it too sensitive?
I tried fix and changed the prompt for 10 times by 2 accounts. And this is what i got in every single time.
My Prompt : “Create a dreamy, poetic video in a vibrant painterly style, inspired by the provided image. Emphasize the serene, ethereal, and atmosphere with soft, atmospheric lighting. Cinematic.”
Hi @Rin_Rin,
Are you also getting issue without image (like with normal prompt for video generation) ??
Hi, i think Veo 2 working perfectly without video generate by image.
Prompt 1: “Generate a short video in a painterly, impressionistic style with visible brushstrokes, capturing the look and feel of the provided image. Show a woman standing on a gentle slope in a field of vibrant yellow and purple wildflowers. She is wearing a hat, a yellow jacket over a light cream dress, and is focused on painting on an easel positioned in front of her. She holds a paintbrush in one hand and a round palette in the other. Large, soft, voluminous clouds drift slowly across a bright, light-blue sky. A gentle breeze causes the flowers and tall grass to sway, sends yellow petals drifting through the air, and makes a ribbon on the easel flutter subtly. Maintain the original artwork’s soft lighting, color palette, and artistic texture throughout the animation. Low angle perspective”
Prompt 2 : “A female artist in a yellow jacket and hat paints en plein air on an easel in a sunlit field brimming with yellow wildflowers. Large, soft, painterly clouds drift across the bright sky. A gentle breeze sways the flowers and lifts yellow petals into the air. Painterly, impressionistic style with a dreamy atmosphere, soft lighting.”
We saw some issue yesterday generating video using image but it got resolved today. Let me check on this and will get back to you if there is still some issue.
I get the same error whenever there are any people in the image, even AI generated people looking up at the clouds in the sky. This wasn’t the case before. I think after the recent upgrade the default “safety settings” have been changed drastically.
Add in the configuration:
““config: {
personGeneration: “allow_adult”,
aspectRatio: “16:9”,
numberOfVideos: 1,
},””,
add personGeneration: “allow_adult”, when the model will generate the video. I tested this model, and found it very expensive, using the api, it is unfeasible for the price I was charged for 128 seconds.
