Lite
Fastest and lightest; good for many tests, drafts, and social-media direction exploration.
Fast
Balances speed and quality, supports 4K, and fits most product demos, short ads, and client previews.
Premium
Standard high-quality mode with 4K support for important shots and delivery versions.
Why Choose Veo 3.1
Veo 3.1 can keep visuals, motion, and synchronized sound effects/dialogue in one generation flow, which works well for product demos, short ads, social assets, and client demo videos.
Start with Lite for quick tests, then choose Fast or Premium after the prompt and visual direction are stable.
Extend video
Videos generated with Fast and Standard/Premium can be extended by 7 seconds each time, up to 20 extensions. From an 8-second source video, the maximum continuous result is 8 + 7 × 20 = 148 seconds.
Lite mode does not support video extension. The video used for extension must have been generated or extended in the last 2 days; extension creates a new merged MP4, keeps the source job, and does not overwrite the original video.
Selling points
Why We Focus On Veo
Clear Delivery Quality
Covers 720p, 1080p, and 4K output. 4K is available for Fast and Premium, which suits ads, ecommerce, and large-screen demo assets.
Vertical First
Supports 9:16 vertical generation for full-screen mobile video on Shorts, Reels, Douyin, Xiaohongshu, and similar channels.
Reference Image Control
Upload up to 3 product, person, or scene reference images so the visual direction, product look, and style are easier to keep consistent.
Visuals And Audio Together
Supports native sound effects, ambience, and dialogue generation, reducing later voiceover, music, and sync work.
Why It Fits Commercial Service
PhotonMark Veo focuses on a deliverable workflow: transparent credits, preview before generation, private MP4 downloads, job history, and automatic refunds on failure.
We provide the service on Google DeepMind Veo 3.1 for fast 4-8 second clip iteration. Actual generation time depends on the model, 4K, audio, reference images, and Google's available capacity. If Vertex AI returns a capacity, safety, or call failure, PhotonMark marks the job as failed and automatically refunds credits instead of leaving credits stuck in an unexplained queue.
We use official Google Vertex AI capabilities and safety policies, not gray-market accounts, reselling shortcuts, or unexplained model access. For brand, ecommerce, and client-demo projects, that is more reliable than exaggerated model comparisons.
CNBC reported in 2025: Google confirmed that it uses a subset of its YouTube video library to train AI models such as Gemini and Veo 3. We treat this as context about Google's video ecosystem scale, not as a promise about specific training material or generated results; commercial delivery still depends on official Vertex AI access, safety policies, and a traceable workflow.