GLASNOST V.2: 80s Soviet Art-Video Collage

Style/Context Low Rank Adaptor (LoRA)
For Wan2.1 14B T2V & I2V Base Models
Stylers of Kinema Historical LoRAs
|||||||| By SilverAgePoets.com ||||||||

Prompt: [GLASNOST] style...

Prompt: [GLASNOST] style...

Prompt: [GLASNOST] style...

Prompt: [GLASNOST] style...

Prompt: [GLASNOST] style...

Prompt: [GLASNOST] style...

About this LoRA

This is a Rank 16/Alpha 64 LoRA for the Wan2.1 14b video generation model.
It may be used to generate several distinct scene-windows-concepts within a single clip (not unlike the well-known ZOOM LoRA).
We've found that given certain prompting styles and LoRA strength modifications may enable controlled gradations of inter-cohesion between the scenes.

It was trained on 100+ manually edited (by us) collages/montages, largely using the same clips and frames used to train the other GLASNOST LoRA (V.1), but with some additions specific to this variant.
These clips & frames were sourced by us from a variety of iconic 1980s Perestroika-era Soviet films, tv shows, concerts, & music videos.
Overall, the sources for this version of GLASNOST lean further into the realm of underground/countercultural/art film territories, with some Leningrad Metamodernist, Moscow Conceptualist, as well as all sorts of Soviet rock influences represented.
The captions this time around should enable this LoRA to exhibit slighly better knowledge (than V.1) of names like Yegor Letov, Viktor Tsoy, Yanka Dyaghileva, or bands Auctyon, KINO, Nol, and others.

This adapter can be used with Wan as well as Skyreels via diffusers or ComfyUI or DrawThings, etc...
This LoRA works well with both CausVid & Self-Forcing distillation quick inference adapters.
It also works fairly well in combos w/ other LoRAs.
Get creative with these!

Trigger words

You should use GLASNOST style Perestroika-era 1980s Soviet detailed experimental arthouse film sequence. On top left is ____ , on top right is ____, on bottom left is ____, on bottom right is ____, video filmed in the USSR during the perestroika era, featuring several concurrent clips of 16mm footage as a thematically-unified cinematographic collage of several distinct scenes, vintage Soviet television, underground cinema, radical metamodernist cinepoetry, from an award-winning real life raw mixed media conceptual sots art video filmed in the USSR during the Perestroika era, Leningrad punk, Moscow conceptualism, etc, to revive one of these more recent gestalts of futures no-longer-past!

Using with Diffusers

pip install git+https://github.com/huggingface/diffusers.git

import torch
from diffusers.utils import export_to_video
from diffusers import AutoencoderKLWan, WanPipeline
from diffusers.schedulers.scheduling_unipc_multistep import UniPCMultistepScheduler

model_id = "wavespeed/Wan2.1-T2V-14B-Diffusers-fp16"
vae = AutoencoderKLWan.from_pretrained(model_id, subfolder="vae", torch_dtype=torch.float32)
pipe = WanPipeline.from_pretrained(model_id, vae=vae, torch_dtype=torch.bfloat16)
flow_shift = 3.0  # 5.0 for 720P, 3.0 for 480P
pipe.scheduler = UniPCMultistepScheduler.from_config(pipe.scheduler.config, flow_shift=flow_shift)
pipe.to("cuda")

pipe.load_lora_weights("AlekseyCalvin/Glasnost_v1_wan_14b_USSR80sTVstyle")

pipe.enable_model_cpu_offload() #for low-vram environments

prompt = "GLASNOST style Perestroika-era 1980s Soviet detailed experimental arthouse film sequence. On top left is ____ , on top right is ____, on bottom left is ____, on bottom right is ____, video filmed in the USSR during the perestroika era, featuring several concurrent clips of 16mm footage as a thematically-unified cinematographic collage of several distinct scenes, vintage Soviet television, underground cinema, radical metamodernist cinepoetry, from an award-winning real life raw mixed media conceptual sots art video filmed in the USSR during the Perestroika era, Leningrad punk, Moscow conceptualism"
negative_prompt = "overexposed, static, blurred, subtitles, images, static, worst, low, JPEG compression residue, incomplete, extra fingers, poorly drawn, poorly drawn, deformed, disfigured, misshapen, fused, still picture, backwards"

output = pipe(
    prompt=prompt,
    negative_prompt=negative_prompt,
    height=480,
    width=832,
    num_frames=81,
    guidance_scale=5.0,
).frames[0]
export_to_video(output, "output.mp4", fps=16)

Training details

Steps: 4000
Learning rate: 0.0002
LoRA rank: 16 dim, 64 alpha

Contribute your own examples

You can use the community tab to add videos that show off what you’ve made with this LoRA.