Essential cookies keep Instavar working. Optional analytics help us understand how the site is used and link your first-visit source to your Studio account after sign-in, for up to 180 days. Cookie Policy

Manage Cookie Preferences

Necessary - authentication, security, and consent storage (always on)Analytics - Google Analytics 4 with IP anonymisation and account-linked first-visit attribution for up to 180 days. Withdraw through Cookie Settings at any time.

Service reliability telemetry, including Sentry error monitoring and Vercel Speed Insights, stays enabled so we can secure the product and diagnose failures.

Skip to content

Research Playbooks Tools Open source Case studies Studio

Home
Blog

Singapore Office - Antiphishing Pte. Ltd.

JTC LaunchPad @ one-north

67 Ayer Rajah Crescent, #02-14

Singapore 139950

© 2026 Instavar. All rights reserved.

Find the angle that travels.

Research Playbooks Tools Open source Case studies Studio About Careers|Privacy Terms CookiesAI Policy Rights & Consent Report Abuse

UK Corporate Presence - Litiga Ltd

Registered in England and Wales

Company number 11610573

Registered office: 128 City Road

London, England EC1V 2NX

Incorporated 8 October 2018

Instavar Blog

Make better content decisions.

Practical lessons on video, distribution, and measurement. Start with the question you need to answer, then go deeper when it helps.

Search posts

Latest notes.

Surya OCR 2 Layout Analysis - Why Higher F1 Did Not Make It Our Default
Surya OCR 2 beat Paddle on layout F1 in two public-safe tests, but weaker label coverage, fragmented boxes, and fallback behavior kept it out of our default OCR pipeline.
How to Position a Kids Furniture Business by Studying What Parents Compare
A concrete method for studying what parents worry about, what furniture brands they compare, what public discussions reveal, and what each website makes easy or hard to picture before writing a clearer positioning line.
How We Used SkillOpt to Stress-Test Codex Skills

We adapted the SkillOpt idea into a practical test loop for Codex skills: synthetic edge cases, narrow edits, and fresh unseen score streaks before we trusted an improvement.

Open-Source Lip Sync Models Compared in 2026

A practical comparison of LatentSync, KeySync, MuseTalk, InfiniteTalk, LTX LipDub, VideoReTalking, Wav2Lip, Diff2Lip, Hallo, EchoMimic, and TalkVerse for production video dubbing workflows.

Supertonic 3 On-Device TTS Reality Check on macOS

A first-party Supertonic 3 smoke test on Apple Silicon macOS: ONNX Runtime CPU setup, model cache size, RAM usage, audio output, no-discrete-VRAM behavior, and where it fits against Qwen3-TTS, Kokoro, F5-TTS, and CosyVoice.

Function Calling and MCP First Principles

A plain-language walkthrough of how function calling and MCP fit together in a production agent loop.

AI Video Anchor Frames: First and Last Frame Continuity Playbook

A practical playbook for using first-frame and last-frame anchors in AI video generation, grounded in a Seedance titration workflow where anchors fixed colour continuity, exposed composition drift, and shaped the final QA gate.

FP8 on RTX 3090 Ti - What Actually Works on Consumer GPUs

FP8 on an RTX 3090 Ti is useful, but not for the reason most examples imply. Ampere SM 8.6 can use FP8 as a storage dtype for VRAM savings, while native FP8 tensor-core compute starts with newer GPU architectures. Here is the practical guide for FLUX-style image and video pipelines on 24 GB cards.

LoRA vs Full SFT for Voice Models - What Actually Changes on a 24 GB GPU

A practical guide to choosing LoRA or full supervised fine-tuning for TTS and voice cloning models on RTX 3090 Ti and RTX 4090 class GPUs, using first-party Qwen3-TTS, VoxCPM, CosyVoice, and IndexTTS2 runs.

Mining Claude Code and Codex Logs Into a Knowledge Base

A practical guide to turning agent transcripts into reusable engineering memory without dumping every JSONL file into a vector database.

VoxCPM 2 Full SFT on a 24 GB GPU - The Run That Actually Fit

First-party notes from a VoxCPM 2 full supervised fine-tuning run on an RTX 3090 Ti, including the memory failures, the working PagedAdamW8bit recipe, dataset cleanup, validation results, and checkpoint selection.

OpenAI Codex CLI /goal Command: What It Does and How to Enable It

A source-inspected guide to the experimental OpenAI Codex CLI /goal command: what it stores, how it lets Codex continue work, how to enable the goals feature today, and why Codex Desktop users should treat it as CLI-only for now.

Page 2 of 11