Vheer AI Review

The homepage, before I signed in

I opened the site without an account first, just to see what I was being promised. A banner ran across the top pushing a new model, and under the hero heading the page told me I could start free with unlimited generations, no signup, and watermark-free exports.

The claim I latched onto was the no signup part, since I ended up making an account anyway before I could reach anything useful. Nothing was broken here. The page loaded quickly and the Start Creating Free button sat where I expected it.

 The dashboard after logging in

Once I logged in I landed on a dashboard with every tool stacked down the left rail and the same tools repeated as cards in the main panel. My balance showed 50 credit points next to my email in the bottom corner.

Two things stood out immediately. The credit counter is small and tucked into the corner where I nearly missed it, and the sidebar duplicates the card grid, so the screen carries the same set of links twice with nothing extra added the second time.

Setting up my first text to image run

I went to Text to Image and pasted in a long prompt about a futuristic warrior on a rain-soaked skyscraper above a cyberpunk city. For the model I picked Vheer Quality, because it was the one that cost nothing. Anything else on that dropdown would have started eating my credits. Aspect ratio 1:1, one image.

The prompt box holds about six lines before it starts scrolling, so most of what I wrote vanished into a scroll well and I could not read it back without dragging through it. What I did appreciate is that the Generate button prints the cost on its own face. Mine read 0, so I knew the run was free before I committed to it.

 The result

About a minute later the image appeared in the preview panel with edit, undo, and download controls under it.

It caught most of what I asked for. The armor came back black with blue LED lines, the cape is caught mid-movement, the helmet is tucked under one arm, and the rooftop is wet enough to throw reflections back at the neon. The face is the weak spot, a little waxy across the cheeks. Every billboard in the background carries invented lettering that reads as nothing at all.

The downloaded file

I downloaded it to see what actually lands on disk. Square, 1248 by 1248, and no watermark anywhere on it, which matched what the homepage advertised.

At full size the armor plating and the falling rain hold up better than I expected from a free tier. The helmet under his arm is the giveaway. Enlarged, it never quite resolves into an object with a front and a back.

 Trying text to video

Next I moved to Text to Video and asked for a giant glowing jellyfish drifting over a futuristic city at sunset. SeeDance V1 Pro Fast, aspect ratio 4:3, duration 4 seconds, resolution 720p.

The Generate button now read 24. Four seconds of video was about to cost 24 of my 50 credits, so I could afford this run and roughly one more before I ran dry. A looping ferris wheel animation played in the preview area while I set everything up, which I briefly mistook for something I had made.

Blocked by the plan

I clicked Generate and a red-bordered warning appeared in the top corner telling me my current plan cannot access this model and that I should upgrade my subscription to use premium models.

This was the most frustrating moment of the whole session. My balance was fine at 50 credits against a stated price of 24. The plan was the thing standing in the way, not the arithmetic. I went back through the model dropdown afterwards and every video model listed there was gated the same way, which makes the price printed on the Generate button actively misleading. Quoting me a credit cost for something I am not permitted to run set me up to fail before I pressed anything.

 Image to image with my own output

I loaded the warrior I had already generated into Image to Image and asked for him to become a medieval knight in ornate silver and gold armor on a castle tower during a thunderstorm. Vheer Quality again, aspect ratio 1:1.

Cost showed 0, so the free model carries across to this tool as well. The panel is laid out identically to Text to Image, which meant there was nothing new to learn before I could use it.

The knight

Roughly another minute of waiting.

The knight came back in engraved silver and gold plate with a crimson cape and a sword glowing orange down the length of the blade, castle towers behind him and lightning splitting the sky on the left. The scene swap is complete. Nothing cyberpunk survived into the frame.

The surprise was the face. It kept the same man from my first image almost exactly, which I never asked it to do.

The downloaded knight

Clean download again, square, same 1248 pixels. Chainmail at the collar, rivets across the pauldrons, gold filigree worked into the breastplate, wet stone underfoot.

The sword is what gives it away at full size. It glows like a layer dropped on top of the picture rather than a light source doing anything to the armor beside it.

 Feeding the knight into image to video

For Image to Video the top of the screen splits into two frame slots, first frame required and end frame optional. I loaded the knight into the first and left the second empty. Vheer Quality, aspect ratio on Auto, duration 5 seconds, resolution 768.

Cost showed 0. That is the part I could not reconcile. Video was locked behind a subscription over in Text to Video, yet the same account on the same plan was about to make a video here for nothing.

 The generated video

Two minutes this time, roughly double what the images took. The clip plays back inside the panel with a Download button under it.

I set the duration to 5 seconds and the player reads 0:04 as the total length, so that setting did not carry through. The motion is cautious. His cape shifts and the rain keeps falling, and the camera pushes in a fraction before settling. The circling hero shot I described in the prompt never happened.

The hairstyle changer

I went into Portrait and Styles for the Hairstyle Changer, loaded the cyberpunk warrior back in, set the style dropdown to Buzz Cut, and typed Change the hairstyle to Buzz Cut into the prompt box. Vheer Quality, aspect ratio 1:1.

Typing that felt redundant. There is a dropdown that already names the style and a prompt box asking for the same instruction in words, and nothing on screen tells me which one takes priority if the two disagree. A GoDaddy ad also sat at the bottom of the settings column, parked directly under the Generate button.

Buzz cut applied

A minute later, the haircut was done.

The rest of the frame survived the edit. Same rooftop, same billboards, same armor colour, same wide stance. Looking at it in the preview I could already tell it had touched more than the head.

 The downloaded version

At full size that became obvious. The framing pulled back, so there is more roof at the bottom edge and more sky above him than the original had. His cape hangs in a different shape. The chest harness picked up extra straps that were not there before, and the helmet under his arm changed shape yet again.

So the tool is not editing hair on top of my picture, it is regenerating the entire image and holding it close to the original. His face came back softer and rounder than the first pass, which is a strange thing to inherit from a request about hair.