Bulk image & video generation on magnific.com, straight from a side panel.
This extension turns a list of prompts into a queue and runs them on magnific.com for you — generating and downloading the results hands-free. You pick your model and any Magnific-side settings (resolution, etc.) on Magnific itself; the extension generates against whatever you've selected there and handles the bulk work. Keep the Magnific tab visible while a batch runs — browsers give inactive tabs fewer resources (not an extension issue), so automation in a background tab may not work correctly. To work on something else meanwhile, keep the tab visible in its own window or on a second monitor.
Magnific Automation is preconfigured browser automation for Magnific. It does not ask ChatGPT, Claude, Codex, Manus, Perplexity, or another general AI agent to inspect the page and decide every click. You provide the prompts and Magnific settings; the extension runs the known Magnific workflow and handles the repetitive execution.
This avoids spending separate LLM-agent credits on repeatable browser work. Magnific generation itself can still use Magnific credits or a paid Magnific plan.
Pick a mode at the top of the Control tab. The extension works on Magnific's image generator page or video generator page depending on the mode. The extension does NOT switch pages for you — open the right Magnific page in your active tab before running.
Prompt → video. Type prompts (one per blank-line-separated block) and the extension generates a video for each on Magnific's video generator page.
Prompt → image. Type prompts and generate images on Magnific's image generator page. Use the extension's Outputs / Prompt dropdown (1 / 2 / 4) to choose how many variants each prompt generates — the extension drives Magnific's own count stepper for you before each submit, so whatever you pick in the side panel wins.
Image + prompt → video. Upload one source image and enter a prompt on Magnific's video generator page. The extension sets the start-frame slot, submits the prompt, and collects the resulting video outputs. If the current Magnific model/settings do not expose a start-frame slot, the extension blocks the run and tells you what to change.
Start frame + end frame + prompt → video. Upload source images, choose Adjacent or Grouped pair logic in the side panel, and run on Magnific's video generator page. This mode requires both visible start and end frame slots; if the active model/settings only support a start frame, use Image → Video instead.
Source images + prompt → image. Upload source images on the extension side panel
and run on Magnific's image generator page. For each job the extension clears the
previous references, uploads the new ones, reads Magnific's visible @imgN
label for each, and includes those tokens when it submits the prompt.
References per prompt (1–6) sets how many images one job may use. With Auto-attach images off, your uploads are grouped in order — with 2 per prompt, images 1+2 go to prompt 1, images 3+4 to prompt 2, and a final odd image still forms a valid job on its own.
Turn Auto-attach images on to match by filename instead: an image is attached to
any prompt that mentions its name (without the extension, ignoring case and punctuation).
blue-wolf.png matches “draw the blue wolf”. Matching is whole-name, so
anna.png will not attach to “annabelle” — but in Korean, Japanese, Chinese and
Thai a particle may follow the name directly, as those languages are written that way.
Very short names are ignored in this mode, because a file called a.png would
otherwise match nearly every prompt.
Type one prompt per line. Blank lines are ignored, so you can space things out while you work. A prompt cannot span several lines — each line becomes its own generation. The status strip below the textarea shows the count ("Ready — 12 text-to-video prompts") so you can confirm what was actually parsed before you spend anything.
You can also upload a .txt file and choose whether it should Append to the current prompts or Replace them. For Google Sheets, Excel, or another spreadsheet, open the spreadsheet importer: it recognises tab-separated or CSV rows and shows exactly how many prompts will be appended or used as a replacement before you apply them.
For long lists, use the prompt navigator above the editor. It shows the current parsed prompt count; enter a number and click Jump to select that prompt without manually scrolling.
For image-driven modes, use the Prompt matching preview before running. It shows each prompt row beside the source image, start frame, or end frame that will be submitted, so you can catch ordering mistakes before credits are used.
Concurrent Prompts (1–5) controls the maximum number of active generated cards/jobs at once while preserving sequence order. If prompt 1 asks for 4 outputs and concurrency is 5, the extension can submit prompt 1 four times and prompt 2 once; when one output finishes, the next ordered output starts.
Higher concurrency = faster batches but more strain on your Magnific account. If your model is rate-limited, set this lower. If Magnific is generous, you can push it up to 5.
Between every submit, the extension waits a random number of seconds in the range you set (default 3–5s). Randomising the delay makes batches look less like a bot to Magnific's rate limiters. Increase it if you see moderation flags or generation throttling.
Magnific's content moderation occasionally false-positives — a benign prompt gets flagged as a policy violation. Retries on Block is enabled by default and can be turned off in Settings. Maximum Retries defaults to 1 and counts retries after the first attempt. A timeout continues watching the same submitted generation instead of spending credits or creating a duplicate by clicking Generate again.
Settings lets you choose the extension language and the generation mode selected when the panel opens. Changing Default Mode also switches the current workflow immediately; choosing a mode tile for one batch does not replace that saved preference. Use Browser settings under Download Settings to open your browser's Downloads controls, where you can choose the browser's default download location and related behavior.
Choose Auto-download in the Control tab's Configuration block before running. When it is on, each finished image/video is saved to your Downloads folder the moment Magnific finishes generating it. The extension clicks Magnific's own download button and applies any selected folder or filename rule through Chrome's download rename flow.
The Save to folder input creates a real subfolder under Downloads (e.g.
Downloads/magnific-batch/). Leave it empty to save into the Downloads root.
The File naming controls let you keep Magnific's native filename or use the prompt
text as the filename. You can also turn on the order prefix
(001_, 001_v1_, …) so files sort by prompt order even when
Magnific finishes them out of order.
Stop = "no more new submits." Prompts that are still waiting in Magnific's queue will continue generating and downloading; prompts that haven't been submitted yet are cancelled. Magnific has no cancel API for in-flight generations — once submitted, the credit is spent regardless of what we do client-side.
After Stop, the prompt queue shows an explicit resume bar with Resume and Start fresh actions. The footer Run button always starts the current form as a fresh batch, so an old resume point cannot accidentally hijack a new test run.
The Run button is always clickable. If you click it while a batch is already running, the new batch is queued and starts as soon as the current batch finishes (including its reconciliation pass). The queue card shows the current batch's progress and any pending batches stacked below.
After every batch, the extension does a final sweep of Magnific's feed to catch any cards that finished generating AFTER our main wait timed out (feed virtualisation can hide freshly-rendered cards from polling). Recovered cards are downloaded and the CSV row is promoted from "failed" to "success (reconciled)". This typically completes within a few seconds; in the worst case it runs for up to 30 seconds.
If Create CSV is on (default), every batch ends with an auto-downloaded CSV containing one row per prompt, with the prompt text, status (success / moderated / timeout / etc.), timestamp, and image/video URLs. Useful for keeping a record of what was generated.
storage (settings sync), tabs (detect the active Magnific tab),
sidePanel, scripting (inject content script into already-open Magnific tabs after install/update),
alarms (keepalive + hourly compatibility refresh), identity (Google sign-in option),
downloads (save generated files into a named subfolder).
The extension couldn't reach magnific-automation.pages.dev. Click Retry;
if it persists, check your network/firewall. The extension genuinely cannot operate without
a successful fetch — see the Requirements section above for why.
The extension auto-injects into already-open Magnific tabs on install/update, but if you opened the tab BEFORE installing, the injection may take a second. Refresh the Magnific tab if the first Run doesn't work.
Check the Debug Logs tab. If you see ✓ Prompt N generated but no
↓ Saved line, the click on Magnific's download button didn't fire a Chrome
download — usually because Magnific's React handler swallowed the click silently. Retry the
prompt, or download manually from Magnific's UI.
001_, pulley got 002_)The download-rename pipeline has a 1-second hold between clicks to serialize Chrome's download queue and prevent prefix-swaps at high concurrency. If you still see swapped prefixes, drop concurrency to 2 or 1 and the issue goes away.
Compatibility updates are delivered through magnific-automation.pages.dev without
a Chrome Web Store republish. If automation breaks suddenly, message me on X
(link below) — a Cloudflare deploy can fix it within minutes and your extension picks up
the fix within an hour (or sooner if you reload the side panel).
The extension stores your settings in chrome.storage.sync (syncs across browsers
signed into the same Google account). Firebase Auth tokens live in IndexedDB on your device.
No prompts, generation URLs, or other usage data leave your browser — everything happens
between your browser, magnific.com, and Cloudflare Pages (for compatibility checks).
Full details in the privacy policy.