DarthAffe/stable-diffusion.cpp
1
0
Fork 0
mirror of https://github.com/leejet/stable-diffusion.cpp.git synced 2026-08-10 06:07:58 +00:00
Code Issues Packages Projects Releases Wiki Activity

feat: move circular padding from context to per-generation params (#1748)

Browse Source
This commit is contained in:
fszontagh 2026-07-06 16:04:20 +02:00 committed by GitHub
parent dff0e88265
commit e071aa390b
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
4 changed files with 53 additions and 42 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

33
examples/common/common.cpp
View File

@ -554,18 +554,6 @@ ArgOptions SDContextParams::get_options() {
"--vae-conv-direct", "--vae-conv-direct",
"use ggml_conv2d_direct in the vae model", "use ggml_conv2d_direct in the vae model",
true, &vae_conv_direct}, true, &vae_conv_direct},
{"",
"--circular",
"enable circular padding for convolutions",
true, &circular},
{"",
"--circularx",
"enable circular RoPE wrapping on x-axis (width) only",
true, &circular_x},
{"",
"--circulary",
"enable circular RoPE wrapping on y-axis (height) only",
true, &circular_y},
{"", {"",
"--chroma-disable-dit-mask", "--chroma-disable-dit-mask",
"disable dit mask for chroma", "disable dit mask for chroma",
@ -853,9 +841,6 @@ std::string SDContextParams::to_string() const {
<< " diffusion_flash_attn: " << (diffusion_flash_attn ? "true" : "false") << ",\n" << " diffusion_flash_attn: " << (diffusion_flash_attn ? "true" : "false") << ",\n"
<< " diffusion_conv_direct: " << (diffusion_conv_direct ? "true" : "false") << ",\n" << " diffusion_conv_direct: " << (diffusion_conv_direct ? "true" : "false") << ",\n"
<< " vae_conv_direct: " << (vae_conv_direct ? "true" : "false") << ",\n" << " vae_conv_direct: " << (vae_conv_direct ? "true" : "false") << ",\n"
<< " circular: " << (circular ? "true" : "false") << ",\n"
<< " circular_x: " << (circular_x ? "true" : "false") << ",\n"
<< " circular_y: " << (circular_y ? "true" : "false") << ",\n"
<< " chroma_use_dit_mask: " << (chroma_use_dit_mask ? "true" : "false") << ",\n" << " chroma_use_dit_mask: " << (chroma_use_dit_mask ? "true" : "false") << ",\n"
<< " qwen_image_zero_cond_t: " << (qwen_image_zero_cond_t ? "true" : "false") << ",\n" << " qwen_image_zero_cond_t: " << (qwen_image_zero_cond_t ? "true" : "false") << ",\n"
<< " chroma_use_t5_mask: " << (chroma_use_t5_mask ? "true" : "false") << ",\n" << " chroma_use_t5_mask: " << (chroma_use_t5_mask ? "true" : "false") << ",\n"
@ -912,8 +897,6 @@ sd_ctx_params_t SDContextParams::to_sd_ctx_params_t(bool taesd_preview) {
sd_ctx_params.tae_preview_only = taesd_preview; sd_ctx_params.tae_preview_only = taesd_preview;
sd_ctx_params.diffusion_conv_direct = diffusion_conv_direct; sd_ctx_params.diffusion_conv_direct = diffusion_conv_direct;
sd_ctx_params.vae_conv_direct = vae_conv_direct; sd_ctx_params.vae_conv_direct = vae_conv_direct;
sd_ctx_params.circular_x = circular || circular_x;
sd_ctx_params.circular_y = circular || circular_y;
sd_ctx_params.force_sdxl_vae_conv_scale = force_sdxl_vae_conv_scale; sd_ctx_params.force_sdxl_vae_conv_scale = force_sdxl_vae_conv_scale;
sd_ctx_params.chroma_use_dit_mask = chroma_use_dit_mask; sd_ctx_params.chroma_use_dit_mask = chroma_use_dit_mask;
sd_ctx_params.chroma_use_t5_mask = chroma_use_t5_mask; sd_ctx_params.chroma_use_t5_mask = chroma_use_t5_mask;
@ -1189,6 +1172,18 @@ ArgOptions SDGenerationParams::get_options() {
"disable auto resize of ref images", "disable auto resize of ref images",
false, false,
&auto_resize_ref_image}, &auto_resize_ref_image},
{"",
"--circular",
"enable circular padding on both axes for tileable output",
true, &circular},
{"",
"--circularx",
"enable circular padding on x-axis (width) only",
true, &circular_x},
{"",
"--circulary",
"enable circular padding on y-axis (height) only",
true, &circular_y},
{"", {"",
"--disable-image-metadata", "--disable-image-metadata",
"do not embed generation metadata on image files", "do not embed generation metadata on image files",
@ -2475,6 +2470,8 @@ sd_img_gen_params_t SDGenerationParams::to_sd_img_gen_params_t() {
params.hires.upscale_tile_size = hires_upscale_tile_size; params.hires.upscale_tile_size = hires_upscale_tile_size;
params.hires.custom_sigmas = hires_custom_sigmas.empty() ? nullptr : hires_custom_sigmas.data(); params.hires.custom_sigmas = hires_custom_sigmas.empty() ? nullptr : hires_custom_sigmas.data();
params.hires.custom_sigmas_count = static_cast<int>(hires_custom_sigmas.size()); params.hires.custom_sigmas_count = static_cast<int>(hires_custom_sigmas.size());
params.circular_x = circular || circular_x;
params.circular_y = circular || circular_y;
return params; return params;
} }
@ -2540,6 +2537,8 @@ sd_vid_gen_params_t SDGenerationParams::to_sd_vid_gen_params_t() {
params.hires.upscale_tile_size = hires_upscale_tile_size; params.hires.upscale_tile_size = hires_upscale_tile_size;
params.hires.custom_sigmas = hires_custom_sigmas.empty() ? nullptr : hires_custom_sigmas.data(); params.hires.custom_sigmas = hires_custom_sigmas.empty() ? nullptr : hires_custom_sigmas.data();
params.hires.custom_sigmas_count = static_cast<int>(hires_custom_sigmas.size()); params.hires.custom_sigmas_count = static_cast<int>(hires_custom_sigmas.size());
params.circular_x = circular || circular_x;
params.circular_y = circular || circular_y;
return params; return params;
} }

8
examples/common/common.h
View File

@ -165,10 +165,6 @@ struct SDContextParams {
bool diffusion_conv_direct = false; bool diffusion_conv_direct = false;
bool vae_conv_direct = false; bool vae_conv_direct = false;
bool circular = false;
bool circular_x = false;
bool circular_y = false;
bool chroma_use_dit_mask = true; bool chroma_use_dit_mask = true;
bool chroma_use_t5_mask = false; bool chroma_use_t5_mask = false;
int chroma_t5_mask_pad = 1; int chroma_t5_mask_pad = 1;
@ -246,6 +242,10 @@ struct SDGenerationParams {
int upscale_repeats = 1; int upscale_repeats = 1;
int upscale_tile_size = 128; int upscale_tile_size = 128;
bool circular = false;
bool circular_x = false;
bool circular_y = false;
bool hires_enabled = false; bool hires_enabled = false;
std::string hires_upscaler = "Latent"; std::string hires_upscaler = "Latent";
std::string hires_upscaler_model_path; std::string hires_upscaler_model_path;

6
include/stable-diffusion.h
View File

@ -216,8 +216,6 @@ typedef struct {
bool tae_preview_only; bool tae_preview_only;
bool diffusion_conv_direct; bool diffusion_conv_direct;
bool vae_conv_direct; bool vae_conv_direct;
bool circular_x;
bool circular_y;
bool force_sdxl_vae_conv_scale; bool force_sdxl_vae_conv_scale;
bool chroma_use_dit_mask; bool chroma_use_dit_mask;
bool chroma_use_t5_mask; bool chroma_use_t5_mask;
@ -385,6 +383,8 @@ typedef struct {
sd_cache_params_t cache; sd_cache_params_t cache;
sd_hires_params_t hires; sd_hires_params_t hires;
int qwen_image_layers; int qwen_image_layers;
bool circular_x;
bool circular_y;
} sd_img_gen_params_t; } sd_img_gen_params_t;
typedef struct { typedef struct {
@ -410,6 +410,8 @@ typedef struct {
sd_tiling_params_t vae_tiling_params; sd_tiling_params_t vae_tiling_params;
sd_cache_params_t cache; sd_cache_params_t cache;
sd_hires_params_t hires; sd_hires_params_t hires;
bool circular_x;
bool circular_y;
} sd_vid_gen_params_t; } sd_vid_gen_params_t;
typedef struct sd_ctx_t sd_ctx_t; typedef struct sd_ctx_t sd_ctx_t;

48
src/stable-diffusion.cpp
View File

@ -863,10 +863,6 @@ public:
use_tae = true; use_tae = true;
} }
if (sd_ctx_params->circular_x || sd_ctx_params->circular_y) {
LOG_INFO("Using circular padding for convolutions");
}
{ {
if (!ensure_backend_pair(SDBackendModule::TE) || if (!ensure_backend_pair(SDBackendModule::TE) ||
!ensure_backend_pair(SDBackendModule::DIFFUSION)) { !ensure_backend_pair(SDBackendModule::DIFFUSION)) {
@ -1369,15 +1365,6 @@ public:
} }
} }
diffusion_model->set_circular_axes(sd_ctx_params->circular_x, sd_ctx_params->circular_y);
if (high_noise_diffusion_model) {
high_noise_diffusion_model->set_circular_axes(sd_ctx_params->circular_x, sd_ctx_params->circular_y);
}
if (control_net) {
control_net->set_circular_axes(sd_ctx_params->circular_x, sd_ctx_params->circular_y);
}
circular_x = sd_ctx_params->circular_x;
circular_y = sd_ctx_params->circular_y;
} }
LOG_DEBUG("validating model metadata"); LOG_DEBUG("validating model metadata");
@ -3061,8 +3048,6 @@ void sd_ctx_params_init(sd_ctx_params_t* sd_ctx_params) {
sd_ctx_params->eager_load = false; sd_ctx_params->eager_load = false;
sd_ctx_params->enable_mmap = false; sd_ctx_params->enable_mmap = false;
sd_ctx_params->diffusion_flash_attn = false; sd_ctx_params->diffusion_flash_attn = false;
sd_ctx_params->circular_x = false;
sd_ctx_params->circular_y = false;
sd_ctx_params->chroma_use_dit_mask = true; sd_ctx_params->chroma_use_dit_mask = true;
sd_ctx_params->chroma_use_t5_mask = false; sd_ctx_params->chroma_use_t5_mask = false;
sd_ctx_params->chroma_t5_mask_pad = 1; sd_ctx_params->chroma_t5_mask_pad = 1;
@ -3114,8 +3099,6 @@ char* sd_ctx_params_to_str(const sd_ctx_params_t* sd_ctx_params) {
"auto_fit: %s\n" "auto_fit: %s\n"
"flash_attn: %s\n" "flash_attn: %s\n"
"diffusion_flash_attn: %s\n" "diffusion_flash_attn: %s\n"
"circular_x: %s\n"
"circular_y: %s\n"
"chroma_use_dit_mask: %s\n" "chroma_use_dit_mask: %s\n"
"chroma_use_t5_mask: %s\n" "chroma_use_t5_mask: %s\n"
"chroma_t5_mask_pad: %d\n" "chroma_t5_mask_pad: %d\n"
@ -3152,8 +3135,6 @@ char* sd_ctx_params_to_str(const sd_ctx_params_t* sd_ctx_params) {
BOOL_STR(sd_ctx_params->auto_fit), BOOL_STR(sd_ctx_params->auto_fit),
BOOL_STR(sd_ctx_params->flash_attn), BOOL_STR(sd_ctx_params->flash_attn),
BOOL_STR(sd_ctx_params->diffusion_flash_attn), BOOL_STR(sd_ctx_params->diffusion_flash_attn),
BOOL_STR(sd_ctx_params->circular_x),
BOOL_STR(sd_ctx_params->circular_y),
BOOL_STR(sd_ctx_params->chroma_use_dit_mask), BOOL_STR(sd_ctx_params->chroma_use_dit_mask),
BOOL_STR(sd_ctx_params->chroma_use_t5_mask), BOOL_STR(sd_ctx_params->chroma_use_t5_mask),
sd_ctx_params->chroma_t5_mask_pad, sd_ctx_params->chroma_t5_mask_pad,
@ -3234,6 +3215,8 @@ void sd_img_gen_params_init(sd_img_gen_params_t* sd_img_gen_params) {
sd_img_gen_params->batch_count = 1; sd_img_gen_params->batch_count = 1;
sd_img_gen_params->control_strength = 0.9f; sd_img_gen_params->control_strength = 0.9f;
sd_img_gen_params->qwen_image_layers = 3; sd_img_gen_params->qwen_image_layers = 3;
sd_img_gen_params->circular_x = false;
sd_img_gen_params->circular_y = false;
sd_img_gen_params->pm_params = {nullptr, 0, nullptr, 20.f}; sd_img_gen_params->pm_params = {nullptr, 0, nullptr, 20.f};
sd_img_gen_params->pulid_params = {nullptr, 1.0f}; sd_img_gen_params->pulid_params = {nullptr, 1.0f};
sd_img_gen_params->vae_tiling_params = {false, false, 0, 0, 0.5f, 0.0f, 0.0f, nullptr}; sd_img_gen_params->vae_tiling_params = {false, false, 0, 0, 0.5f, 0.0f, 0.0f, nullptr};
@ -3267,6 +3250,8 @@ char* sd_img_gen_params_to_str(const sd_img_gen_params_t* sd_img_gen_params) {
"control_strength: %.2f\n" "control_strength: %.2f\n"
"photo maker: {style_strength = %.2f, id_images_count = %d, id_embed_path = %s}\n" "photo maker: {style_strength = %.2f, id_images_count = %d, id_embed_path = %s}\n"
"VAE tiling: %s (temporal=%s, extra_tiling_args=%s)\n" "VAE tiling: %s (temporal=%s, extra_tiling_args=%s)\n"
"circular_x: %s\n"
"circular_y: %s\n"
"hires: {enabled=%s, upscaler=%s, model_path=%s, scale=%.2f, target=%dx%d, steps=%d, denoising_strength=%.2f}\n", "hires: {enabled=%s, upscaler=%s, model_path=%s, scale=%.2f, target=%dx%d, steps=%d, denoising_strength=%.2f}\n",
SAFE_STR(sd_img_gen_params->prompt), SAFE_STR(sd_img_gen_params->prompt),
SAFE_STR(sd_img_gen_params->negative_prompt), SAFE_STR(sd_img_gen_params->negative_prompt),
@ -3288,6 +3273,8 @@ char* sd_img_gen_params_to_str(const sd_img_gen_params_t* sd_img_gen_params) {
BOOL_STR(sd_img_gen_params->vae_tiling_params.enabled), BOOL_STR(sd_img_gen_params->vae_tiling_params.enabled),
BOOL_STR(sd_img_gen_params->vae_tiling_params.temporal_tiling), BOOL_STR(sd_img_gen_params->vae_tiling_params.temporal_tiling),
SAFE_STR(sd_img_gen_params->vae_tiling_params.extra_tiling_args), SAFE_STR(sd_img_gen_params->vae_tiling_params.extra_tiling_args),
BOOL_STR(sd_img_gen_params->circular_x),
BOOL_STR(sd_img_gen_params->circular_y),
BOOL_STR(sd_img_gen_params->hires.enabled), BOOL_STR(sd_img_gen_params->hires.enabled),
sd_hires_upscaler_name(sd_img_gen_params->hires.upscaler), sd_hires_upscaler_name(sd_img_gen_params->hires.upscaler),
SAFE_STR(sd_img_gen_params->hires.model_path), SAFE_STR(sd_img_gen_params->hires.model_path),
@ -3336,6 +3323,8 @@ void sd_vid_gen_params_init(sd_vid_gen_params_t* sd_vid_gen_params) {
sd_vid_gen_params->hires.upscale_tile_size = 128; sd_vid_gen_params->hires.upscale_tile_size = 128;
sd_vid_gen_params->hires.custom_sigmas = nullptr; sd_vid_gen_params->hires.custom_sigmas = nullptr;
sd_vid_gen_params->hires.custom_sigmas_count = 0; sd_vid_gen_params->hires.custom_sigmas_count = 0;
sd_vid_gen_params->circular_x = false;
sd_vid_gen_params->circular_y = false;
sd_cache_params_init(&sd_vid_gen_params->cache); sd_cache_params_init(&sd_vid_gen_params->cache);
} }
@ -4211,6 +4200,25 @@ struct CircularAxesState {
bool circular_y = false; bool circular_y = false;
}; };
static void apply_circular_axes_to_diffusion(sd_ctx_t* sd_ctx, bool circular_x, bool circular_y) {
sd_ctx->sd->circular_x = circular_x;
sd_ctx->sd->circular_y = circular_y;
if (sd_ctx->sd->diffusion_model) {
sd_ctx->sd->diffusion_model->set_circular_axes(circular_x, circular_y);
}
if (sd_ctx->sd->high_noise_diffusion_model) {
sd_ctx->sd->high_noise_diffusion_model->set_circular_axes(circular_x, circular_y);
}
if (sd_ctx->sd->control_net) {
sd_ctx->sd->control_net->set_circular_axes(circular_x, circular_y);
}
if (circular_x || circular_y) {
LOG_INFO("Using circular padding for convolutions (x=%s, y=%s)",
circular_x ? "true" : "false",
circular_y ? "true" : "false");
}
}
static CircularAxesState configure_image_vae_axes(sd_ctx_t* sd_ctx, static CircularAxesState configure_image_vae_axes(sd_ctx_t* sd_ctx,
const sd_img_gen_params_t* sd_img_gen_params, const sd_img_gen_params_t* sd_img_gen_params,
const GenerationRequest& request) { const GenerationRequest& request) {
@ -4967,6 +4975,7 @@ SD_API bool generate_image(sd_ctx_t* sd_ctx,
sd_ctx->sd->sampler_rng->manual_seed(request.seed); sd_ctx->sd->sampler_rng->manual_seed(request.seed);
sd_ctx->sd->set_flow_shift(sd_img_gen_params->sample_params.flow_shift); sd_ctx->sd->set_flow_shift(sd_img_gen_params->sample_params.flow_shift);
sd_ctx->sd->apply_loras(sd_img_gen_params->loras, sd_img_gen_params->lora_count); sd_ctx->sd->apply_loras(sd_img_gen_params->loras, sd_img_gen_params->lora_count);
apply_circular_axes_to_diffusion(sd_ctx, sd_img_gen_params->circular_x, sd_img_gen_params->circular_y);
ImageVaeAxesGuard axes_guard(sd_ctx, sd_img_gen_params, request); ImageVaeAxesGuard axes_guard(sd_ctx, sd_img_gen_params, request);
@ -5805,6 +5814,7 @@ SD_API bool generate_video(sd_ctx_t* sd_ctx,
} }
int64_t t0 = ggml_time_ms(); int64_t t0 = ggml_time_ms();
sd_ctx->sd->vae_tiling_params = sd_vid_gen_params->vae_tiling_params; sd_ctx->sd->vae_tiling_params = sd_vid_gen_params->vae_tiling_params;
apply_circular_axes_to_diffusion(sd_ctx, sd_vid_gen_params->circular_x, sd_vid_gen_params->circular_y);
GenerationRequest request(sd_ctx, sd_vid_gen_params); GenerationRequest request(sd_ctx, sd_vid_gen_params);
bool latent_upscale_enabled = request.hires.enabled; bool latent_upscale_enabled = request.hires.enabled;
GenerationRequest hires_request = request; GenerationRequest hires_request = request;