Zhenzhen's AI Radar · MiniMax H3 · 2026-11-13 Evening Report
This issue features 3 new versions: Strongly Recommended 0 · Recommended 0 · Worth Watching 2 · Included Only 1
Editor's Note This issue reviewed 3 new versions of H3, with 2 rated B and 1 rated C; there were no S or A ratings. The top recommendation is the volumetric fog slider from the KinetiX Weather series: positive values spread ground-hugging fog in mid-to-far distances, while negative values remove fog. The author provided tiered weights, but the model is in paid early access and all 9 examples are hidden. Unblinking prevents characters from blinking in videos, facilitating frame extraction for character references; it is also paid and has no visible examples. The Movement Amplitude Slider v2 is permanently paid, lacks details on differences from v1, and has 10 examples with 9 hidden, resulting in a downgrade from B to C.
Worth Watching (2)
[B] KinetiX Weather System Series · Volumetric Fog v1.0
LoRA · Version Update · Author make676767 · Published 11-13 17:05 · Downloads 3 · Paid
Volumetric fog slider LoRA from the KinetiX Weather series: positive values spread thin ground mist up to dense fog in mid-to-far distances while keeping the foreground clear; negative values remove haze for clear distant views.
Image

Reasons for Recommendation
- Series update: This is the 4th version of the KinetiX Weather System Series, but it is not a revision of old weights; rather, it introduces a new effect dimension (previous versions covered rain, day/night, and snow); the version notes only excerpt the description for the highest tier.
- Atmosphere control for text-to-video / image-to-video: The author claims it only adds fog without adding sunbeams or washing out the overall image; example scenes include night alleys, rainy nights, and dawn forests.
- The author provided tiered weights: +0.8 to +1.5 for slight airiness, +2.0 to +3.0 for recommended cinematic morning fog, +4.0 to +5.0 for dense fog engulfing mid-to-far distances, and -2.0 to -5.0 for fog removal.
How to Use
- No trigger words required; load the LoRA and set intensity by tier, with the author recommending +2.0 to +3.0; when adding fog, use prompts like 'dense volumetric fog, swirling ground mist', and for fog removal, use 'crystal clear air, zero fog'.
- The author's two finished templates: ground-hugging morning fog in an ancient forest uses +2.5, and cold fog in a cyberpunk rainy alley uses +2.2; the author did not specify the number of steps.
Risks and Limitations
- Paid early access, with the API indicating it ends on 2026-11-16.
- All 9 examples are hidden, so actual effects cannot be seen; the author recommends combining it with the rain slider from the same series for rainy night cold fog, but there are no examples for the combined effect.
- The description only includes weights and prompt templates, with no information on training data or compatible resolutions/durations.
- The card image is an AI-generated illustration, not the model's actual output; the author has no publicly available standard-level examples.
[B] Unblinking - MiniMax H3 Lora (Surpress Blinking Eyes) · v1.0
LoRA · New Model · Author saliacoel · Published 11-13 09:01 · Downloads 0 · Paid
Concept LoRA for suppressing blinking: keeps characters' eyes open and prevents blinking throughout the entire video.
Image

Reasons for Recommendation
- New model addressing a specific problem: generated characters blink naturally, making it easy to capture closed-eye frames when extracting frames for character reference images or when a specific frame result is needed; the author states it is used for creating character reference images and videos requiring specific frame results.
- The author provided trigger words and intensity, making usage simple.
How to Use
- Add 'UN8L1Nk1NG.' and 'Eyes stay static open.' to the prompt; intensity is listed as approximately 0.75 in the version notes and approximately 1.0 in the model card.
- The author did not specify the number of steps or accompanying nodes.
Risks and Limitations
- Paid early access, with the API indicating it ends on 2026-11-16; there were 0 downloads at the time of release.
- The only example is hidden, with no visible examples, making it impossible to judge the impact on expressions or other eye movements.
- The description is only a few lines, lacking details on training data and applicable scenarios.
- The card image is an AI-generated illustration, not the model's actual output; the author has no publicly available standard-level examples.
Trigger Words: UN8L1Nk1NG, Eyes stay static open.
Included Only (1)
- Movement Amplitude Slider MMH3 · v2.0 — LoRA: Movement Amplitude Slider LoRA v2: positive values amplify limb extension, weapon swing arcs, and jump spans, while negative values tighten movements into close-quarters combat.

Data: Civitai Public API (base model MiniMax H3), inclusion window 11-13 08:31 → 11-13 21:01 (Beijing Time). Recommendation reasons have been reviewed by editors; download counts only reflect attention levels.