Heavy Flowing Breasts V2 is an Illustrious LoRA focused on giving breasts a stronger sense of weight, softness, gravity, and natural drape rather than simply making them larger.
It emphasizes lower fullness, subtle overhang, soft deformation, body contact, and a less spherical silhouette. The goal is heavy, not saggy — full and substantial, but still retaining structure rather than becoming excessively drooped or pendulous.
Trigger: hfbshape
The trigger is recommended for consistency, but early testing suggests V2 can express the learned shape even without it when the LoRA is loaded.
Recommended starting strength: 0.2–0.6
V2 can already produce a noticeable effect at relatively low strengths. Higher values strengthen the learned morphology, but may also increase breast size, chest emphasis, or other secondary effects.
Works best with: medium-to-large breasts, especially when the pose allows their weight, lower contour, drape, compression, or body contact to remain visible.
Less effective with: small breasts and strongly lifted/perky breast shapes, where there is less volume available for the characteristic shape change to show clearly.
V2 is intended primarily as a shape LoRA rather than a breast-size LoRA, although some size increase can still occur — particularly on medium breasts. The training set leaned toward larger breasts, so a mild size bias remains.
Early testing also suggests V2 is considerably less prone than V1 to overriding the original camera angle, framing, or composition.
Pose and clothing behavior are still being tested. Clothed results so far are promising, although the LoRA may encourage stronger chest emphasis, cleavage, or exposure even at relatively low strengths.
Description
V2 is a full retrain focused on better isolating the intended breast morphology from unrelated training biases.
Introduces the dedicated trigger
hfbshape.Retrained on a more carefully curated and varied dataset with cleaner, more consistent captions.
Greater variation in pose, camera angle, framing, and subject composition was used to reduce the perspective and composition bias present in V1.
Early testing shows substantially less tendency to override camera angle, framing, or overall composition.
Improved emphasis on weight, lower fullness, subtle overhang, soft deformation, and natural breast-to-body interaction.
Still primarily intended as a shape LoRA, though some breast-size increase can occur. Medium breasts in particular may become somewhat larger due to the larger-breasted training set.
Pose and clothing performance are still being tested and this information may be updated as testing continues.