[Platform][OpenRouter] Introduce video generation

[lochmueller]

Q	A
Bug fix?	no
New feature?	yes
Docs?	yes
Issues
License	MIT

Add the new video generation API of OpenRouter incl. example.

Docs: https://openrouter.ai/docs/guides/overview/multimodal/video-generation

[chr-hertel]

quite an assumption that it is first_frame and not last_frame ... 🤔

[chr-hertel]

What keeps me a bit busy thinking here is the input data - what makes sense?

for text-to-speech i just merged a plain string - that's what we also usually use if there's only string as input - no need for wrapping it into Text - but then there's that frame_images (even with frame_type), input_references, and provider.

what do you think about making them options?

$result = $platform->invoke('google/veo-3.1-lite', 'Animate the blue elephant, [
    'duration' => 4,
    'resolution' => '720p',
    'aspect_ratio' => '16:9',
    'generate_audio' => false,
    'first_frame' => new ImageUrl(...),
    'last_frame' => Image::fromFile(...),
    'input_reference => new ImageUrl(...),
    'provider' => [
        'options' => [
            'google-vertex' => [
                'parameters' [
                    'personGeneration' => 'allow',
                    'negativePrompt' => 'blurry, low quality',
                ],
            ],
        ],
      ],
    ],
]);

[chr-hertel]

With PR #2408 merged you could rebase and adopt async jobs here :)

[chr-hertel]

Sorry, this comment still needs to be addressed:
#1975 (comment)