A 2D Transformer model for image-like data. Parameters: num_attention_heads (`int`, *optional*, defaults to 16): The number of heads

(convolution => [BN] => ReLU) * 2

Borrowed from https://github.com/facebookresearch/pytorch3d Notice: x,y,z are in image space, normalized can only render squared

Alg: https://www.scratchapixel.com/lessons/3d-basic-rendering/rasterization-practical-implementation Notice: x,y,z are in image space, no

r""" A basic Transformer block. Parameters: dim (`int`): The number of channels in the input and output. num_attention_heads

borrowed from https://github.com/soubhiksanyal/FLAME_PyTorch/blob/master/FLAME.py Given flame parameters this class generates a differentiabl

FLAME texture: https://github.com/TimoBolkart/TF_FLAME/blob/ade0ab152300ec5f0e8555d6765411555c5ed43d/sample_texture.py#L64 FLAME texture

The output of [`Transformer2DModel`]. Args: sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)` or `(ba

The output of [`UNet2DConditionModel`]. Args: sample (`torch.FloatTensor` of shape `(batch_size, num_channels, height, width)`):

A 2D UNet mid-block [`UNetMidBlock2D`] with multiple residual blocks and optional attention blocks. Args: in_channels (`int`): The n

Main Class

Tiny Autoencoder block used in [`AutoencoderTiny`]. It is a mini residual module consisting of plain conv + ReLU blocks. Args: i

Downscaling with maxpool then double conv

r""" A conditional 2D UNet model that takes a noisy sample, conditional state, and a timestep and returns a sample shaped output. This mo

Upscaling then double conv