Gemini Omni 1.2 Flash: The Good, the Bad and the Ugly | 34 AI Clips, Same Prompts

Google has announced Gemini Omni 1.2 Flash: 15-second clips, more consistent scenes and a 30% lower price. But is it actually better? We took 34 AI video clips we made with Omni 1.1 Flash for this channel, regenerated them with the exact same prompts in the new model, and put them side by side. Here's the good, the bad and the ugly of Omni 1.2 Flash. In this video: ✅ What's new in Omni 1.2 Flash: 15-second clips, consistency, references, conversational editing ✅ Google Flow vs. Google Vids: credits or seconds? (500 seconds per month on a Pro plan) ✅ Old vs. new model: same prompt, 34 side-by-side comparisons ✅ The prompt the new model refused, the quota limit and the Gemini logo Where do we use these AI visuals? ▶ Horizon Europe Essentials | Episode 1:    • Horizon Europe Essentials: How the €93.5 B...   Note: Omni 1.2 Flash isn't in Google Flow yet, so we generated the clips in Google Vids. Chapters 0:00 Omni 1.2 Flash is out 0:33 What's new? 1:45 How we use these visuals 2:35 Flow vs. Vids: credits or seconds? 3:35 First comparison: city after a quake 4:02 Findings: the good, the bad and the ugly 7:10 The comparisons (34 clips) 7:27 Farm robot: from lab to orchard 7:43 Autonomous litter-collecting catamaran 8:00 Smart solar glass for a hospital 8:16 Hydrogen city bus 8:32 Supercomputer and big-science infrastructure 8:48 Researcher mobility (MSCA) 9:04 A consortium for global challenges 9:20 Investor pitch and thought bubble 9:36 From satellite to drone 9:52 3D blueprint: scanning a city after a quake 10:08 Isometric diorama: seismic observation 10:24 Four hazard cards 10:40 3D blueprint: drone modules 10:56 Point cloud: before and after 11:12 Damage classification screen 11:28 Handover from university to SME 11:44 Physical drone and digital twin 12:00 Harsh conditions: night, dust and wind 12:16 AI that works without internet 12:32 Scanning for voids under rubble 12:48 Race: manual inspection vs. the system 13:04 Isometric diorama: city drill 13:20 Three drones, one neighbourhood 13:36 Field test during an aftershock 13:52 Digital twin: flood, fire and quake 14:08 3D blueprint: damage verification 14:24 Public procurement and crisis systems 14:40 Crisis-centre integration 14:56 The tree that grows again 15:12 Full operation in a cascading disaster 15:28 Mediterranean resilience network 15:44 A bridge over the valley of death (refused) 16:01 Isometric diorama: canyon bridge 16:17 3D blueprint: funding gates 16:33 Consortium summit in Brussels (quota reached) 16:49 Wrap-up: next, character consistency If you liked the video, hit like and Hype, and subscribe. Character consistency is our next test! #GeminiOmni #AIVideo #GoogleAI