这是一个3D奶油CG风格的画风lora，有Z image base版本和LTX2.3版本，如果你想让LTX2.3呈现最好的效果，请使用Z image base lora 生成图片，再用LTX2.3模型加上风格lora图生视频，下面给出视频示例参考提示词：
(场景): 暖光温馨客厅，灰色沙发，阳光透过窗户洒入，氛围治愈松弛。
(角色): 美女，黑长直齐刘海，身穿白衬衫 + 灰百褶裙，抱着木吉他坐在沙发上，身后趴着一只软萌柯基。
(分镜 1), 中景固定机位：少女坐在沙发上，低头专注弹吉他，柯基趴在她身后，耳朵跟着旋律轻轻晃动。
(分镜 2), 近景慢镜头：她的手指在琴弦上流畅拨动，阳光在吉他上反射出柔和光斑，柯基凑过来蹭她的肩膀。
(分镜 3), 中景推近：少女停下弹奏，低头看向柯基，笑着 say，“你也喜欢这首歌吗？”
(分镜 4), 中景定格：她轻轻摸了摸柯基的头，画面在暖光中缓缓收尾。
This is a LoRA with a 3D creamy CG art style, available in Z Image Base version and LTX2.3 version. If you want LTX2.3 to deliver the best results, generate images with the Z Image Base LoRA first, then use the LTX2.3 model together with the style LoRA to create videos from images. Below are reference prompt examples for the video:
(Scene): A warm and cozy living room with a grey sofa, sunlight streaming through the window, creating a healing and relaxed atmosphere.
(Character): A beautiful young woman with long straight black hair and straight bangs, wearing a white shirt and a grey pleated skirt. She sits on the sofa holding an acoustic guitar, with an adorable Corgi lying prone behind her.
(Shot 1), Medium shot fixed camera: The girl sits on the sofa, bowing her head to play the guitar intently. The Corgi lies behind her, its ears swaying gently to the melody.
(Shot 2), Close-up slow motion: Her fingers pluck the guitar strings smoothly. Soft light spots reflect on the guitar in the sunlight, and the Corgi comes over to nuzzle her shoulder.
(Shot 3), Medium shot push-in: The girl stops playing, looks down at the Corgi and smiles, saying, "Do you like this song too?"
(Shot 4), Medium shot freeze frame: She gently pats the Corgi's head, and the scene fades out slowly in warm light.
Description
无需触发词，可以添加常用的动漫类质量词语获得更好的效果，生成图片后，有对应的LTX2.3版本画风Lora用来图生视频。
No trigger words are required. You can add commonly used anime-style quality keywords for better results. After generating images, there is a corresponding Lora with the LTX2.3 style available for image-to-video generation.
FAQ
Comments (9)
very nice work
Is it possible for you to add the link to this creamy 3D CG art style LoRA for Zimage Base?
This link contains two versions of LoRA.
This has been great for keeping the faces of my generations beautiful without morphing them. Love it, Thank you ❤️
Pretty impressive consistency for LTX.
Thank you for sharing! Your example videos look impressive. They have one of the smoothest animations I've seen for a 2.5d anime style. What did you train the lora on? How many videos? Images? Can you share more information about your training so we can learn from you?
This LoRA was trained on 198 images. Honestly, the results didn’t live up to my expectations. It only picked up half of the target style. Since it’s an image-based training run, all the motion and movement still come from the base LTX 2.3 model itself.
Luckily, using image-to-video yields great results.
@liao942287300dog It is rather impressive for an image only training run. Your images must have hit a sweet spot, making it correlate them with 3d CGI animation instead of shitty stiff 2D animation, causing them to start rendering proper animation.
Honestly, I always use img2vid instead of txt2vid. It's is faster and easier to just generate an image with a txt2img model, fix whatever you want fixed on it and then use it for video then wait minutes for the video to render and hope it will magically come out exactly as you had hoped. (It wont.)