The lesson
Publish a model benchmark that names the rivals, explains the scoring rule in one line, lets readers inspect the outputs themselves, and says where a competitor still wins.
A 'RESEARCH' label sits over the title, followed by an intro that promises 'how we tested it in blind comparisons' and 'where it still falls short', then a shaded 'Key Takeaways' box. 'How we evaluated JST-2 Motion' names the rivals and the prompt buckets, and explains that 'Score = win rate + ½ draw rate' and that 50% means equal preference. The 'Text-to-motion' and 'Video-to-motion' sections each open with a 'DRAG TO ORBIT' 3D viewer and a row of clips such as 'Dodge and duck' and 'Pirouette', then give per-rival bar charts split into wins, draws and rival wins for each criterion. 'What comes next' lists the remaining limits, including that 'Fast sports and athletics footage is where Cartwheel Comic 4 still leads.'
Don't copy Viggle's criteria, rival list or scores, or the MetaHuman rig setup. Keep the order of takeaways, method, inspectable outputs, per-rival charts with draws shown, and an honest 'what comes next' section.
Tell us what you want to make. We’ll talk through how to put these ideas to work for your brand.