Fine Tuning Friday - Qwen & WAN 2.2 - Multi Character Video Generation
- Date
- 2025-09-19
- Host
- Oxen.ai
About this event
Fine Tuning Friday is a hands-on meetup for people who want to get sharper at modern model customization, not just talk about it in the abstract. This session focuses on Qwen and WAN 2.2 in the context of multi-character video generation—a topic that sits right at the edge of where creative tooling, workflow design, and practical fine-tuning start to matter. If you care about making video outputs more consistent, more controllable, and more usable across characters and scenes, this is the kind of room you want to be in. Expect a technical, community-driven conversation with enough structure to learn something real and enough openness to compare notes with other builders, experimenters, and curious practitioners. About the Event Fine Tuning Friday is built around a simple idea: bring people together regularly to explore how fine-tuning and model adaptation actually work in practice. Rather than staying at the level of broad AI headlines, this meetup narrows in on a concrete workflow area and gives attendees space to dig into the methods, tradeoffs, and implementation questions behind it. This edition centers on Qwen & WAN 2.2 and how they relate to multi-character video generation. That means the conversation is likely to matter to people thinking about character consistency, prompt reliability, training data choices, model behavior across scenes, and the broader challenge of getting generative systems to produce outputs that are not just impressive once, but repeatable. Because this is part of a recurring community format, the event is also designed to be social and conversational. You are not showing up for a distant, one-way presentation. You are showing up to learn with other people who are testing tools, comparing results, and trying to close the gap between what models can do in demos and what they can do in real workflows. Whether you are already running experiments or just starting to understand how fine-tuning fits into video generation, the format makes room for both technical curiosity and practical discussion. The value comes from seeing how others are approaching similar problems and hearing how they think through constraints, failure cases, and opportunities. What to Expect The session will revolve around the topic of multi-character video generation, with discussion anchored in the tooling and model context named in the event title. While the exact flow may vary, you should expect a meetup rhythm that combines focused discussion with informal peer exchange. A typical experience at an event like this may include: A clear framing of the day’s topic and why it matters Discussion of fine-tuning concepts relevant to Qwen and WAN 2.2 Exploration of workflow challenges specific to multi-character generation Community Q&A, idea-sharing, and practical back-and-forth Time to connect with other attendees working on adjacent problems The strongest sessions in this format usually come from specificity. That means conversations may get into questions like dataset strategy, consistency across outputs, prompting versus fine-tuning, evaluation of results, and how to think about quality when more than one character has to remain coherent over time. Because the event is in person, expect the social side to be a real part of the experience rather than an afterthought. Side conversations, quick troubleshooting, tool recommendations, and spontaneous whiteboard-style exchanges are often where attendees get the most useful insight. Why Attend If you have been trying to move from general AI interest into sharper technical understanding, this event gives you a focused entry point. The topic is narrow enough to be useful and broad enough to spark ideas across video, creative tooling, research experimentation, and applied generative workflows. You should come if you want to better understand how people are thinking about controllability in generative video. Multi-character work introduces complexity fast: identity drift, scene coherence, role separation, and output stability all become much harder once you move beyond a single subject. A meetup centered on these issues can save you time by exposing what others are learning in real time. There is also value in the room itself. Events tagged as community, networking, meetup, and social tend to attract people who are not only interested in the ideas, but actively working through them. That creates better conversations: less abstract hype, more grounded perspective. You may leave with: A clearer mental model of where fine-tuning fits in a video generation pipeline Better questions to ask in your own experiments New ways to think about consistency and control across multiple characters Peer connections with people exploring similar tools and challenges A stronger sense of what is possible now versus what still requires workarounds Practical Details This is an in-person event taking place on Friday, September 19 at 10:00 AM PDT. Being there physically matters: the format is well suited to live discussion, quick feedback loops, and the kind of informal conversation that is hard to recreate online. If you plan to attend, come ready for a session that is likely to be both technical and conversational. You do not need to arrive with all the answers, but it will help to come with a few concrete questions, examples, or areas of curiosity—especially if you are already thinking about fine-tuning, video generation, or multi-character consistency. A few good ways to prepare: Review your current understanding of fine-tuning workflows Think about where multi-character generation breaks down in practice Bring notes on tools, datasets, prompts, or experiments you want to discuss Be ready to meet people and compare approaches openly This is a strong fit for attendees who like learning in public, thinking through technical detail with peers, and leaving with both insight and momentum. If that sounds like how you work best, Fine Tuning Friday should be worth your time.
Who should attend
This will feel especially useful if you like technical meetups with a practical angle and want to talk with other people who are actively exploring generative workflows. - You are experimenting with **fine-tuning** and want a more concrete understanding of how it applies to real output quality, control, and repeatability. - You are working with or curious about **Qwen**, **WAN 2.2**, or adjacent model stacks and want to hear how others are approaching them in practice. - You care about **video generation**, especially the added difficulty of keeping **multiple characters** consistent across scenes, roles, and outputs. - You build creative tools, prototypes, research projects, or internal workflows and want ideas you can actually test after the event. - You learn best by talking through tradeoffs with other practitioners rather than only consuming polished tutorials or high-level content. - You want the mix of **community, networking, and technical discussion** that comes from being in a room with people who are asking similar questions from different angles. If you are somewhere between curious beginner and active builder, this is likely to be a strong fit. You do not need to be an expert to belong here, but you should be excited to engage with the topic seriously.