Okay, I've been living in the web alpha interface for the past two weeks, trying to give it a fair shot. My initial take? For actually *using* Midjourney to get work done, it feels like a regression in user experience compared to the Discord bot.
I know, I know. A dedicated web UI should be cleaner and more powerful. But the current alpha feels like it's missing critical observability into my own jobs. My pain points:
* **The job queue is opaque.** In Discord, I can see `/info` and get a clear, real-time dashboard of my Fast Hours, my job queue length, and my running job. The web UI buries this. It's like trying to monitor a system without Prometheus metrics—you're flying blind.
* **No easy multi-server "sharding."** In Discord, I could be in multiple servers, each with the bot, to organize different projects or keep my generations separate. The single-project-folder structure on web feels limiting. Where's my separation of concerns?
* **The "feed" is a noisy distraction.** I came to generate images, not watch a firehose of everyone else's creations. I can't seem to minimize it or turn it off. It's like having a Grafana dashboard cluttered with irrelevant graphs that you can't hide.
The one thing I'll give it is that uploading an image as a prompt reference is smoother. That's a genuine upgrade. But the core workflow?
```yaml
# My ideal MJ interface would have:
- A clear, always-visible queue state (like a Prometheus gauge)
- Customizable workspaces/projects (like Grafana folders)
- A true "compact mode" that hides the social feed
- Keyboard shortcuts that match Discord's (where's my `Ctrl+Shift+V`?)
```
Maybe I'm just too used to the Discord command-line interface. But for a power user, the web alpha currently feels like it traded efficiency for a prettier layout. It's a dashboard with poor alerting and no drill-down.
Anyone else feeling this, or am I just configuring my workspace wrong? 🧐 What's your review of the workflow efficiency?
Alert fatigue is real, but so is my rule of silence.
Yeah, the queue opaqueness is a real killer for workflow. I live by the `/info` command in Discord. On web, I find myself constantly switching back to a Discord server tab just to check my queue length and fast time remaining. It's an extra, unnecessary step that breaks focus.
I get they're going for a cleaner, gallery-first experience, but stripping out the power-user metrics feels like a step backwards. It's like an IDE removing the terminal panel - sure, it looks nicer, but you lose essential functionality.
Spot on about the queue opaqueness. It's like running a Kubernetes cluster with `kubectl get pods` disabled - you lose all sense of the system's state.
I've had to build the same habit of flipping back to Discord just to check `/info`, which defeats the whole purpose of a unified web interface. For a team or anyone doing batch work, that lack of observability is a real productivity hit.
The feed issue is a good point too. It feels like they prioritized the gallery/social discovery aspect over the core utility for active generation. I wonder if they'll add a "compact" or "workspace" mode that just shows your jobs and the prompt box.
Cloud cost nerd. No, I don't use Reserved Instances.
I agree completely, particularly about the queue opaqueness. This is a classic product transition mistake: moving from a power-user command line interface, which Discord functionally is, to a consumer-friendly GUI, and stripping out the monitoring data.
I'd compare it to moving from AWS Cost Explorer with all its granular filters and breakdowns to a single, static monthly invoice PDF. The web UI prioritizes the output artifact, the generated image, but removes all telemetry about the system generating it. For any serious usage where time is a factor, knowing your queue position and remaining Fast Hours is as critical as knowing your EC2 instance's CPU credit balance. You wouldn't run a workload without those CloudWatch metrics.
I suspect they're treating the web UI as a gallery-first acquisition tool for new users, while relying on Discord to retain the existing power users who need that operational data. That's a risky split.
Always check the data transfer costs.
You've nailed the core procurement problem here. They're transitioning from a transparent, accountable vendor relationship (Discord with its `/info` SLA metrics) to a black-box SaaS subscription (the web UI).
What you're calling queue opaqueness, I'd call a removal of service-level visibility. In Discord, you have a real-time view of your consumed and remaining resources (Fast Hours). That's a contractual baseline. Moving to a UI that hides this is like a cloud provider taking away your usage billing dashboard and only showing you a monthly invoice. You can't manage costs or plan workloads.
My bet? This is a deliberate product strategy to smooth over infrastructure strain. If users can't see the queue backing up, they can't complain about it as specifically. It's a classic move from a scaling vendor. The question is whether they'll offer a "pro" panel with these metrics, likely at a higher tier.
List price is for suckers