Midjourney on X

Praise names the look rather than the usefulness: the best-looking output, "my favorite here", styles people love at a glance. One person rates the aesthetics "unbeatable" and posts the prompt; another says the texture is still in a class of its own; another credits it with coarse 3D depth. A new Thinking Mode, which re-checks V8.2 images for misplaced objects and text and redraws them, is welcomed as fewer wasted attempts on text banners.

The pipeline users are practical: Midjourney for all image generation, with Adobe for planning and editing; Midjourney for the base image, then Grok Imagine or Seedance for motion; prompts drafted in Claude first; a chocolate campaign designed in Midjourney and finished with video and voice tools. A V8 write-up frames Midjourney as the aesthetics-first tool, now faster and better at text.

The doubters are few and brief. One calls it "so annoying". Another misses the old version, saying prompting is ignored more and artists go unmentioned, and is "seriously considering other options". One rates it highly but finds the environment closed; the V8 write-up notes premium modes cost four times the credits.

Missed asks are concrete. Asked to curl a cat character's front paws to its chest, someone got four images back in the original crossed pose; asking for a raised paw instead worked. Another could get a good starting frame but not further keyframes with the same character, and thought style or character training would be needed. A third says prompting is ignored far more than before and gives "meh results".

The look complaint is about a signature that reads as a limit: one calls out "Midjourney skin", plastic and overly smoothed under studio lighting; another says that after enough use you can always tell it is "a midjourneyish" output; a third says images look like what the tool produced in 2022.

On access, the website is called "hella vibecoded", and one person describes helping a 74-year-old change a subscription, slowly noting what each button does. Another calls the environment somewhat closed. On price, one refuses to pay extra for MCP; a V8 write-up says high-end modes, style references and moodboards burn four times the credits and run four times slower.

Against ChatGPT the split is aesthetic versus convenient: one makes art in Midjourney first and then brings it to GPT, rating another model best for straight generating; another is coming back to Midjourney after a spell on ChatGPT because "the images need too much post-processing"; a third keeps GPT for holding a character's details. A ratings post scores Midjourney 8.7/10 for artistic looks, ChatGPT for making and tweaking in one chat, and Ideogram for readable text on posters.

Grok Imagine is used alongside, not instead: Midjourney for the base and the look, Grok Imagine for variations and motion, one pairing Midjourney stills with a video tool for clips. A side-by-side of Grok Imagine against V8.2 drew the most reach of any comparison.

Ming-Image is tested on one prompt twice (2 people): Ming wins on design production — Chinese titles, strict layout, transparent backgrounds, editable layers — while Midjourney is called a strong inspiration renderer. 2 more say generators like this undercut stock photo libraries. One each name Nano Banana on speed and cost, open local models over a "rental", Muse for the familiar flavour with free redraws, and Krea 3's aesthetic.

The API and MCP ask is short and direct: "can we please have API", "Where do I sign?", "I need MCP", "Any news about MCP?" One person lists Midjourney alongside another tool in a tier list of things to build with, saying the only thing missing is an API or MCP because both are otherwise very strong; another has applied to the MCP creative programme and still asks for an API.

The workflow questions are about assembly steps, not features: whether a dragon in an ocean-to-sky image was a separate generation or one prompt, how a tool called Paper fits into someone's Midjourney workflow, and whether a glowing mushroom texture was fixed at the Midjourney 8.2 stage or added later once the shot was animated. The people asking are usually trying to reproduce a look or a pipeline they have seen posted.

The API and MCP wish is usually about not building around a gap: one person says they had to write "my own janky Claude script" to do it and would love to be part of a proper version; another wants to generate in Midjourney's distinctive style inside Claude, which they use every day; another ranks it among their favourite tools to build with and says an API or MCP is the one thing missing.

On control, someone wanted a cat character to change pose and got the original crossed pose in all four images; what worked was describing one limb moving — a front paw raised to shoulder height — rather than a whole-body curl. Another asks whether a glow texture was locked at the image stage or added after animation.

Single voices ask for an app first, for something more open and convenient, and for a graphic novel version of a sci-fi novel.