Merge pull request #134 from faceshiftlabs/fix/background-power

Fixes bug when using both ms-ssim and background power

CHANGELOG.md

@@ -8,6 +8,15 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 ### In Progress
 - [Freezeable layers (encoder/decoder/etc.)](https://github.com/faceshiftlabs/DeepFaceLab/tree/feature/freezable-weights)
 
+## [1.4.1] - 2020-03-24
+### Fixed
+- When both Background Power and MS-SSIM were enabled, the src and dst losses were being overwritten with the 
+  "background power" losses. Fixed so "background power" losses are properly added with the total losses.
+    - *Note: since all the other losses were being skipped when ms-ssim and background loss were being enabled, this had 
+      the side-effect of lowering the memory requirements (and raising the max batch size). With this fix, you may 
+      experience an OOM error on models ran with both these features enabled. I may revisit this in another feature, 
+      allowing you to manually disable certain loss calculations, for similar performance benefits.*
+
 ## [1.4.0] - 2020-03-24
 ### Added
 - [MS-SSIM loss training option](doc/features/ms-ssim)
@@ -60,7 +69,8 @@ This should help with rough areas directly next to the mask
 - Reset stale master branch to [seranus/DeepFaceLab](https://github.com/seranus/DeepFaceLab), 
   21 commits ahead of [iperov/DeepFaceLab](https://github.com/iperov/DeepFaceLab) ([compare](https://github.com/iperov/DeepFaceLab/compare/4818183...seranus:3f5ae05))
 
-[Unreleased]: https://github.com/olivierlacan/keep-a-changelog/compare/v1.4.0...HEAD
+[Unreleased]: https://github.com/olivierlacan/keep-a-changelog/compare/v1.4.1...HEAD
+[1.4.1]: https://github.com/faceshiftlabs/DeepFaceLab/compare/v1.4.0...v1.4.1
 [1.4.0]: https://github.com/faceshiftlabs/DeepFaceLab/compare/v1.3.0...v1.4.0
 [1.3.0]: https://github.com/faceshiftlabs/DeepFaceLab/compare/v1.2.1...v1.3.0
 [1.2.1]: https://github.com/faceshiftlabs/DeepFaceLab/compare/v1.2.0...v1.2.1

models/Model_SAEHD/Model.py

@@ -476,7 +476,7 @@ Examples: df, liae, df-d, df-ud, liae-ud, ...
                     if self.options['background_power'] > 0:
                         bg_factor = self.options['background_power']
                         if self.options['ms_ssim_loss']:
-                            gpu_src_loss = 10 * nn.MsSsim(resolution)(gpu_target_src, gpu_pred_src_src, max_val=1.0)
+                            gpu_src_loss += bg_factor * 10 * nn.MsSsim(resolution)(gpu_target_src, gpu_pred_src_src, max_val=1.0)
                         else:
                             if resolution < 256:
                                 gpu_src_loss +=  bg_factor * tf.reduce_mean ( 10*nn.dssim(gpu_target_src, gpu_pred_src_src, max_val=1.0, filter_size=int(resolution/11.6)), axis=[1])
@@ -518,7 +518,7 @@ Examples: df, liae, df-d, df-ud, liae-ud, ...
                     if self.options['background_power'] > 0:
                         bg_factor = self.options['background_power']
                         if self.options['ms_ssim_loss']:
-                            gpu_src_loss = 10 * nn.MsSsim(resolution)(gpu_target_dst, gpu_pred_dst_dst, max_val=1.0)
+                            gpu_src_loss += bg_factor * 10 * nn.MsSsim(resolution)(gpu_target_dst, gpu_pred_dst_dst, max_val=1.0)
                         else:
                             if resolution < 256:
                                 gpu_dst_loss +=  bg_factor * tf.reduce_mean ( 10*nn.dssim(gpu_target_dst, gpu_pred_dst_dst, max_val=1.0, filter_size=int(resolution/11.6)), axis=[1])