StoryboardHandler (#37): breaks the project's script into a numbered scene-by-scene breakdown (description + image prompt per scene) via OpenAI in JSON mode (added json_mode support to OpenAiDriver for reliable structured output), stored as a JSON asset. VideoGenerationHandler (#38): generates one still image per storyboard scene via StabilityAI — the scene *visuals*, not motion/video clips (no text-to-video model integrated; see the task comment for why). VideoEditingHandler (#39): the step that actually produces a video file — assembles the scene stills into an MP4 via ffmpeg's concat demuxer, timed to the voiceover audio's duration (probed via ffprobe) or a fixed per-scene duration if there's no audio, then muxes the audio track in. Verified against the real ffmpeg/ffprobe binaries on this box, not just Process::fake(). Asset previews (#45, #46): a reusable partial renders each asset inline by type — image/audio/video get real playback elements, text/ JSON gets a truncated <pre> block, everything else a download link. Wired into both the awaiting_approval step view (so approving/ rejecting is an actual content review, not just a word count) and a general project assets grid. Also fixes a latent bug in AssetStorage::url(), found while wiring up previews: it called Storage::url() on the default 'local' disk, which has no configured URL driver and throws in production — only worked before because Storage::fake() is more permissive than the real local driver. Now serves assets through a new authenticated streaming route instead (VideoProjectAssetController). Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
This commit is contained in:
m105
@@ -0,0 +1,116 @@
|
||||
<?php
|
||||
|
||||
namespace App\Support\VideoWorkflow\StepHandlers;
|
||||
|
||||
use App\Models\AiProvider;
|
||||
use App\Models\WorkflowStep;
|
||||
use App\Support\AiProviders\AiGenerationException;
|
||||
use App\Support\AiProviders\AiProviderManager;
|
||||
use App\Support\AssetStorage\AssetStorage;
|
||||
use App\Support\VideoWorkflow\WorkflowEngine;
|
||||
|
||||
/**
|
||||
* Breaks the project's script into a numbered scene-by-scene storyboard —
|
||||
* each scene gets a description and an image-generation prompt — via an
|
||||
* OpenAI-backed AiProvider in JSON mode. Produces the text/description
|
||||
* layer only; VideoGenerationHandler turns each scene's visual_prompt into
|
||||
* an actual image.
|
||||
*/
|
||||
class StoryboardHandler implements StepHandler
|
||||
{
|
||||
public function __construct(
|
||||
private readonly AiProviderManager $aiProviders,
|
||||
private readonly AssetStorage $assets,
|
||||
) {}
|
||||
|
||||
public function type(): string
|
||||
{
|
||||
return 'storyboard';
|
||||
}
|
||||
|
||||
public function handle(WorkflowStep $step, WorkflowEngine $engine): void
|
||||
{
|
||||
$provider = AiProvider::query()->where('provider', 'openai')->where('is_active', true)->first();
|
||||
|
||||
if (! $provider) {
|
||||
$engine->failStep($step, 'No active OpenAI provider configured.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$script = $this->resolveScript($step);
|
||||
|
||||
if (! $script) {
|
||||
$engine->failStep($step, 'No script available to storyboard — complete a script_generation step first.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
try {
|
||||
$result = $this->aiProviders->generate($provider, $this->buildPrompt($step, $script), ['json_mode' => true]);
|
||||
} catch (AiGenerationException $e) {
|
||||
$engine->failStep($step, $e->getMessage());
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$decoded = json_decode($result->content, true);
|
||||
$scenes = $decoded['scenes'] ?? null;
|
||||
|
||||
if (! is_array($scenes) || empty($scenes)) {
|
||||
$engine->failStep($step, 'AI provider returned an unusable storyboard: no scenes found.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$asset = $this->assets->store(
|
||||
$step->videoProject,
|
||||
'other',
|
||||
json_encode(['scenes' => $scenes], JSON_PRETTY_PRINT),
|
||||
'json',
|
||||
step: $step,
|
||||
mimeType: 'application/json',
|
||||
metadata: ['kind' => 'storyboard', 'scene_count' => count($scenes)],
|
||||
);
|
||||
|
||||
$engine->completeStep($step, [
|
||||
'asset_id' => $asset->id,
|
||||
'scene_count' => count($scenes),
|
||||
]);
|
||||
}
|
||||
|
||||
private function resolveScript(WorkflowStep $step): ?string
|
||||
{
|
||||
$input = $step->input ?? [];
|
||||
|
||||
if (! empty($input['script'])) {
|
||||
return $input['script'];
|
||||
}
|
||||
|
||||
$scriptAsset = $step->videoProject->assets()
|
||||
->where('type', 'script')
|
||||
->orderByDesc('version')
|
||||
->first();
|
||||
|
||||
return $scriptAsset ? $this->assets->contents($scriptAsset) : null;
|
||||
}
|
||||
|
||||
private function buildPrompt(WorkflowStep $step, string $script): string
|
||||
{
|
||||
$input = $step->input ?? [];
|
||||
|
||||
$prompt = "Break the following video script into a numbered scene-by-scene storyboard.\n\n"
|
||||
."Script:\n{$script}\n\n"
|
||||
.'For each scene provide: scene_number (integer), description (what happens on screen), '
|
||||
."and visual_prompt (a concise, self-contained image-generation prompt describing this scene's visual).";
|
||||
|
||||
if (! empty($input['style'])) {
|
||||
$prompt .= " Every visual_prompt must consistently reflect this visual style: {$input['style']}.";
|
||||
}
|
||||
|
||||
$prompt .= "\n\nRespond with JSON only, in exactly this shape: "
|
||||
.'{"scenes": [{"scene_number": 1, "description": "...", "visual_prompt": "..."}]}';
|
||||
|
||||
return $prompt;
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,185 @@
|
||||
<?php
|
||||
|
||||
namespace App\Support\VideoWorkflow\StepHandlers;
|
||||
|
||||
use App\Models\Asset;
|
||||
use App\Models\WorkflowStep;
|
||||
use App\Support\AssetStorage\AssetStorage;
|
||||
use App\Support\VideoWorkflow\WorkflowEngine;
|
||||
use Illuminate\Support\Collection;
|
||||
use Illuminate\Support\Facades\File;
|
||||
use Illuminate\Support\Facades\Process;
|
||||
use Illuminate\Support\Facades\Storage;
|
||||
|
||||
/**
|
||||
* Assembles the scene images from a completed video_generation step and
|
||||
* (optionally) the project's voiceover audio into an actual video file,
|
||||
* via the ffmpeg/ffprobe CLIs. Each scene image is shown for an equal
|
||||
* share of the audio's duration (or a fixed duration if there's no
|
||||
* audio); no transitions/effects — that's future scope, not what this
|
||||
* pass covers.
|
||||
*/
|
||||
class VideoEditingHandler implements StepHandler
|
||||
{
|
||||
private const DEFAULT_SECONDS_PER_SCENE = 4.0;
|
||||
|
||||
public function __construct(private readonly AssetStorage $assets) {}
|
||||
|
||||
public function type(): string
|
||||
{
|
||||
return 'video_editing';
|
||||
}
|
||||
|
||||
public function handle(WorkflowStep $step, WorkflowEngine $engine): void
|
||||
{
|
||||
$images = $this->resolveSceneImages($step);
|
||||
|
||||
if (! $images || $images->isEmpty()) {
|
||||
$engine->failStep($step, 'No scene images available — complete a video_generation step first.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$audioAsset = $this->resolveAudio($step);
|
||||
|
||||
$tempDir = storage_path('app/tmp/video-editing-'.$step->id.'-'.uniqid());
|
||||
File::ensureDirectoryExists($tempDir);
|
||||
|
||||
try {
|
||||
$imagePaths = $images->map(fn (Asset $a) => Storage::path($a->disk_path))->all();
|
||||
$audioPath = $audioAsset ? Storage::path($audioAsset->disk_path) : null;
|
||||
|
||||
$duration = null;
|
||||
|
||||
if ($audioPath) {
|
||||
$duration = $this->probeDuration($audioPath);
|
||||
|
||||
if ($duration === null) {
|
||||
$engine->failStep($step, 'Could not read the voiceover audio\'s duration.');
|
||||
|
||||
return;
|
||||
}
|
||||
}
|
||||
|
||||
$input = $step->input ?? [];
|
||||
$perScene = $input['seconds_per_scene'] ?? ($duration ? $duration / count($imagePaths) : self::DEFAULT_SECONDS_PER_SCENE);
|
||||
|
||||
$concatListPath = "{$tempDir}/concat.txt";
|
||||
$this->writeConcatList($concatListPath, $imagePaths, $perScene);
|
||||
|
||||
$silentVideoPath = "{$tempDir}/silent.mp4";
|
||||
$result = Process::timeout(300)->run([
|
||||
'ffmpeg', '-y', '-f', 'concat', '-safe', '0', '-i', $concatListPath,
|
||||
'-vsync', 'vfr', '-pix_fmt', 'yuv420p', $silentVideoPath,
|
||||
]);
|
||||
|
||||
if ($result->failed()) {
|
||||
$engine->failStep($step, 'ffmpeg failed to assemble scene images: '.trim($result->errorOutput()));
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$finalVideoPath = "{$tempDir}/final.mp4";
|
||||
|
||||
if ($audioPath) {
|
||||
$result = Process::timeout(300)->run([
|
||||
'ffmpeg', '-y', '-i', $silentVideoPath, '-i', $audioPath,
|
||||
'-c:v', 'copy', '-c:a', 'aac', '-shortest', $finalVideoPath,
|
||||
]);
|
||||
|
||||
if ($result->failed()) {
|
||||
$engine->failStep($step, 'ffmpeg failed to mux the voiceover audio: '.trim($result->errorOutput()));
|
||||
|
||||
return;
|
||||
}
|
||||
} else {
|
||||
File::copy($silentVideoPath, $finalVideoPath);
|
||||
}
|
||||
|
||||
$asset = $this->assets->store(
|
||||
$step->videoProject,
|
||||
'video',
|
||||
File::get($finalVideoPath),
|
||||
'mp4',
|
||||
step: $step,
|
||||
mimeType: 'video/mp4',
|
||||
metadata: ['scene_count' => count($imagePaths), 'duration_seconds' => $duration ?? ($perScene * count($imagePaths))],
|
||||
);
|
||||
|
||||
$engine->completeStep($step, ['asset_id' => $asset->id]);
|
||||
} finally {
|
||||
File::deleteDirectory($tempDir);
|
||||
}
|
||||
}
|
||||
|
||||
private function probeDuration(string $path): ?float
|
||||
{
|
||||
$result = Process::timeout(60)->run(['ffprobe', '-v', 'error', '-show_entries', 'format=duration', '-of', 'csv=p=0', $path]);
|
||||
|
||||
if ($result->failed()) {
|
||||
return null;
|
||||
}
|
||||
|
||||
$value = trim($result->output());
|
||||
|
||||
return is_numeric($value) ? (float) $value : null;
|
||||
}
|
||||
|
||||
private function writeConcatList(string $path, array $imagePaths, float $secondsPerImage): void
|
||||
{
|
||||
$lines = [];
|
||||
|
||||
foreach ($imagePaths as $imagePath) {
|
||||
$lines[] = "file '".addslashes($imagePath)."'";
|
||||
$lines[] = "duration {$secondsPerImage}";
|
||||
}
|
||||
|
||||
// ffmpeg's concat demuxer ignores the last entry's duration, so the
|
||||
// final image needs to be repeated once more without one.
|
||||
$lines[] = "file '".addslashes(end($imagePaths))."'";
|
||||
|
||||
File::put($path, implode("\n", $lines));
|
||||
}
|
||||
|
||||
/**
|
||||
* @return Collection<int, Asset>|null
|
||||
*/
|
||||
private function resolveSceneImages(WorkflowStep $step): ?Collection
|
||||
{
|
||||
$input = $step->input ?? [];
|
||||
|
||||
$ids = $input['image_asset_ids'] ?? null;
|
||||
|
||||
if (! $ids) {
|
||||
$videoGenStep = $step->videoProject->steps()
|
||||
->where('type', 'video_generation')
|
||||
->where('status', 'completed')
|
||||
->orderByDesc('completed_at')
|
||||
->first();
|
||||
|
||||
$ids = $videoGenStep?->output['asset_ids'] ?? null;
|
||||
}
|
||||
|
||||
if (! $ids) {
|
||||
return null;
|
||||
}
|
||||
|
||||
$assetsById = Asset::whereIn('id', $ids)->get()->keyBy('id');
|
||||
|
||||
return collect($ids)->map(fn ($id) => $assetsById->get($id))->filter()->values();
|
||||
}
|
||||
|
||||
private function resolveAudio(WorkflowStep $step): ?Asset
|
||||
{
|
||||
$input = $step->input ?? [];
|
||||
|
||||
if (! empty($input['audio_asset_id'])) {
|
||||
return Asset::find($input['audio_asset_id']);
|
||||
}
|
||||
|
||||
return $step->videoProject->assets()
|
||||
->where('type', 'audio')
|
||||
->orderByDesc('version')
|
||||
->first();
|
||||
}
|
||||
}
|
||||
@@ -0,0 +1,123 @@
|
||||
<?php
|
||||
|
||||
namespace App\Support\VideoWorkflow\StepHandlers;
|
||||
|
||||
use App\Models\AiProvider;
|
||||
use App\Models\WorkflowStep;
|
||||
use App\Support\AiProviders\AiGenerationException;
|
||||
use App\Support\AiProviders\AiProviderManager;
|
||||
use App\Support\AssetStorage\AssetStorage;
|
||||
use App\Support\VideoWorkflow\WorkflowEngine;
|
||||
|
||||
/**
|
||||
* Generates the scene visuals for a project's storyboard, one still image
|
||||
* per scene, via a StabilityAI-backed AiProvider. This produces the visual
|
||||
* assets VideoEditingHandler later assembles into an actual video file —
|
||||
* it does not itself produce a video (no motion/text-to-video model is
|
||||
* integrated).
|
||||
*/
|
||||
class VideoGenerationHandler implements StepHandler
|
||||
{
|
||||
public function __construct(
|
||||
private readonly AiProviderManager $aiProviders,
|
||||
private readonly AssetStorage $assets,
|
||||
) {}
|
||||
|
||||
public function type(): string
|
||||
{
|
||||
return 'video_generation';
|
||||
}
|
||||
|
||||
public function handle(WorkflowStep $step, WorkflowEngine $engine): void
|
||||
{
|
||||
$provider = AiProvider::query()->where('provider', 'stability_ai')->where('is_active', true)->first();
|
||||
|
||||
if (! $provider) {
|
||||
$engine->failStep($step, 'No active StabilityAI provider configured.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$scenes = $this->resolveScenes($step);
|
||||
|
||||
if (! $scenes) {
|
||||
$engine->failStep($step, 'No storyboard available — complete a storyboard step first.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$style = ($step->input ?? [])['style'] ?? null;
|
||||
$assetIds = [];
|
||||
|
||||
foreach ($scenes as $scene) {
|
||||
$prompt = $scene['visual_prompt'] ?? $scene['description'] ?? null;
|
||||
|
||||
if (! $prompt) {
|
||||
continue;
|
||||
}
|
||||
|
||||
if ($style) {
|
||||
$prompt .= ", {$style}";
|
||||
}
|
||||
|
||||
try {
|
||||
$result = $this->aiProviders->generate($provider, $prompt);
|
||||
} catch (AiGenerationException $e) {
|
||||
$engine->failStep($step, "scene {$scene['scene_number']}: {$e->getMessage()}");
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$extension = str($result->contentType)->after('image/')->toString() ?: 'png';
|
||||
|
||||
$asset = $this->assets->store(
|
||||
$step->videoProject,
|
||||
'image',
|
||||
$result->content,
|
||||
$extension,
|
||||
step: $step,
|
||||
mimeType: $result->contentType,
|
||||
metadata: ['scene_number' => $scene['scene_number'] ?? null, 'prompt' => $prompt],
|
||||
);
|
||||
|
||||
$assetIds[] = $asset->id;
|
||||
}
|
||||
|
||||
if (empty($assetIds)) {
|
||||
$engine->failStep($step, 'No scenes had a usable visual_prompt.');
|
||||
|
||||
return;
|
||||
}
|
||||
|
||||
$engine->completeStep($step, [
|
||||
'asset_ids' => $assetIds,
|
||||
'scene_count' => count($assetIds),
|
||||
]);
|
||||
}
|
||||
|
||||
/**
|
||||
* @return array<int, array<string, mixed>>|null
|
||||
*/
|
||||
private function resolveScenes(WorkflowStep $step): ?array
|
||||
{
|
||||
$input = $step->input ?? [];
|
||||
|
||||
if (! empty($input['scenes'])) {
|
||||
return $input['scenes'];
|
||||
}
|
||||
|
||||
$storyboardAsset = $step->videoProject->assets()
|
||||
->where('type', 'other')
|
||||
->whereJsonContains('metadata->kind', 'storyboard')
|
||||
->orderByDesc('version')
|
||||
->first();
|
||||
|
||||
if (! $storyboardAsset) {
|
||||
return null;
|
||||
}
|
||||
|
||||
$decoded = json_decode($this->assets->contents($storyboardAsset), true);
|
||||
|
||||
return $decoded['scenes'] ?? null;
|
||||
}
|
||||
}
|
||||
Reference in New Issue
Block a user