A journal for people who create
Ideas, new models and changes in your creative studio.
The new tool plans stories, prepares the cast, and illustrates chapters. Seven workflow steps, review checks, export options, and the costs listed in the November 3 announcement.
ReadHiggsfield’s October 31 guide covers building scenes from references and restyling existing footage into anime with Genjutsu.
ReadHiggsfield explains choosing a UGC or Motion template, adding product references and reviewing the output, including packaging, labels and colors.
ReadAn October 29 article introduces AnySpeechRecognizer with a familiar Swift API, engine choice and speaker diarization, alongside AnyDecisionModel.
ReadNVIDIA describes an open-weight model for transcription, speaker attribution, translation and audio questions, guided by natural-language instructions.
ReadAn October 28 guide demonstrates scene building, 3D assets and video generation from Blender’s viewport in a step-by-step workflow.
ReadLiquid AI published the LFM3 family and practical material on adapting models to a specific task.
ReadAi2 released a study and benchmark examining how consistently language models judge the novelty of research ideas.
ReadA look at OpenAI’s September 8 announcement. Two API models offer different speed and editing-precision tradeoffs; this is a provider update.
ReadIBM’s October 21 release adds language prompting and up to 2,000 keywords for transcription. Russian is absent from the published ASR language list.
ReadGoogle DeepMind’s October 6 model card describes image generation and editing, while flagging small-text and character-consistency limitations.
ReadAnnounced by ElevenLabs on October 15, with the article updated October 20. Effects follow video cues and remain on separate layers.
Read
