中文描述

固定侧视镜头、完整的双人格斗画面，以及细腻的90年代街机像素质感。

两名格斗角色在同一条横向地面上对峙、出拳、踢击、跳跃或释放能量招式。阶梯状轮廓、清晰的像素色块和分层舞台背景，让画面呈现经典街机格斗游戏的感觉。重点是能看清动作、人物全身与对战空间的游戏画面。

角色均为按18–24岁设计的虚构成年男性。素材采用40种人物设定，变化肤色、发型、运动型体态和格斗服装，包括背心、训练夹克、道服、运动短裤、长裤、拳套与护腕等组合。同一套像素风格中保留不同的人物外观与左右攻守安排。

每张图片都带有街机对战UI：与角色对应的男性头像、1P/2P标记、左右血条、中央计时器，以及底部能量槽。人物、舞台与界面共同构成完整的对战画面。

素材覆盖街头、码头、霓虹巷道、屋顶、车站、仓库、道场、寺院庭院、海滨、雪景、运河和工业舞台等20种场景设定，包含待机对峙、防御、拳击、踢击、跳跃交锋与特殊招式。

训练素材共100张，附100份对应英文描述，已逐张检查人物、动作、全身构图与UI，并按实际可见内容修正描述。

触发词：sideviewarcadefight

据@Grockster 指出，该LoRA调整到2以上方可起到作用，但示例图是在模型 0.85的环境下生成，请各位在使用时注意调整权重。感谢各位对LoRA提出宝贵意见！

English Description

A side-view fighting-game visual direction with full-body two-player matches and finely detailed 1990s arcade pixel art.

Two fighters share a horizontal fighting plane as they square off, punch, kick, jump or release energy attacks. Stepped contours, readable pixel clusters and layered stage scenery create the feel of a classic arcade match. The focus is on clear actions, complete character silhouettes and readable fighting space.

All characters are fictional adult men designed as ages 18–24. The material uses 40 character design specifications with varied skin tones, hairstyles, athletic builds and fighting outfits. Clothing and equipment include tanks, training jackets, martial-arts uniforms, sports shorts, trousers, boxing gloves and wrist wraps. Character appearances and left/right attack roles vary within a consistent pixel-art direction.

Every image includes an arcade HUD: male portraits corresponding to the fighters, 1P/2P labels, mirrored health bars, a central round timer and energy meters along the bottom. Characters, stages and interface elements form a complete fighting-game frame.

The material spans 20 stage concepts, including streets, cargo docks, neon alleys, rooftops, stations, warehouses, dojos, temple courtyards, seaside locations, snowy towns, canals and industrial settings. Actions include ready stances, defense, punches, kicks, aerial exchanges and special moves.

The training dataset contains 100 individually reviewed synthetic images with 100 matching English captions. Character appearance, actions, full-body framing and HUD elements were checked, and captions were adjusted to the visible content.

Trigger: sideviewarcadefight

According to @Grockster, this LoRA requires a weight setting of above 2 to be effective; however, the example images were generated with a weight of 0.85, so please adjust the weight accordingly when using it. Thank you all for your valuable feedback on this LoRA!