Image Compressor in the browser is not a generate
/free-tools/image-compressor never spends a credit. Do not pay a model to do an in-browser file job.
Image Compressor never spends a credit. It is a Canvas job on the file you already have: JPEG, PNG, or WebP, a live quality preview, a before-and-after size readout. Nothing is uploaded. Nothing is sent to a model. Do not pay an image model to "make this smaller."
Versely has no free tier for generation. Every catalog generate costs credits. That is why this tool exists as a separate surface: it is free because it costs nothing to run.
Shrinking a file is not redrawing it
A compressor re-encodes pixels you already approved. A generate invents new pixels. If the brief is "this hero is 4.2 MB and the landing page will not take it," you do not need Ideogram, Flux, or a 2K restyle. You need a quality slider and a download.
Paying a model to restyle the still "so it compresses better" is how you get a different still. Labels drift. Skin changes. The file gets smaller because it became a different picture. The compressor keeps the picture and trades bits.
Use a model when you do not have the still. Use this tool when you have the still and it is too heavy.
What the slider is actually doing
The page is honest about the trade. Drag quality down until the preview stops looking acceptable, then stop. The size number updates as you go. For photographs, quality around 0.7–0.8 usually removes more than half the file with no difference most viewers would notice. Flat artwork, screenshots, and anything with text degrade earlier: JPEG-style compression puts artifacts on hard edges. Judge the preview, not the number.
Default output is a lossy encode you can see. PNG as a target is lossless: the quality slider does nothing, and re-encoding a PNG as PNG can add overhead. To shrink a PNG meaningfully, change format. WebP is usually far smaller for the same image, with transparency intact. JPEG when something downstream refuses WebP. The compressor will tell you if the output came out larger than the original. Believe it.
No account. No watermark. Disconnect the network and it still runs — there is no upload to miss.
Credits are for a different meter
What it costs is the credit-formula hub: per second, per 1,000 characters, per megapixel, per export. None of those formulas apply here, because no model is called. The cheapest generate in the catalog is still a generate. This page is a 0-credit file job.
A useful split for a production week:
- Compressor: the still is done; the file is fat.
- Image model: the still does not exist, or one locked region has to change.
- Image-to-video: the still is done and you now want motion.
Running the third or the second to solve the first is a billing mistake, not a quality choice.
When not to use it
Do not compress as a substitute for cropping to the ratio you actually need. Do not compress a vector lockup you should keep as a vector. Do not compress, upload to a generate, and compress again — you are stacking artifacts and still paying for the middle step. If the preview looks like mud at 0.7, you needed a different master, not a smaller mud file.
FAQ
Does this spend credits if I am logged in?
No. The tool never calls a model. Login state is irrelevant. There is no credit quote because there is no charge.
Will you see the image?
No. The file is read into memory on your machine, processed there, and offered back as a download. There is no server component for the file. Offline still works.
Why did my PNG get bigger?
PNG is lossless. Re-encoding without a format change cannot do what the slider implies. Switch to WebP to keep transparency and actually lose weight, or keep the original PNG.
Should I compress before I generate video from the still?
Only if a pipeline cap requires it. Image-to-video wants the cleanest still you have. Compress for delivery of the still. Generate from the master. Do not pay a model to compress, and do not pre-destroy the plate you are about to animate.