Drop PL<2 support

README.md

@@ -2,7 +2,7 @@
 
 | **PyPI** | **Versions** | **Docs & License** | **Tests** | **Reference** |
 |:---------:|:-------------:|:------------------:|:----------:|:--------------:|
-| [![PyPI](https://img.shields.io/pypi/v/opensr-srgan)](https://pypi.org/project/opensr-srgan/) | ![PythonVersion](https://img.shields.io/badge/Python-v3.10%20v3.12-blue.svg)<br>![PLVersion](https://img.shields.io/badge/PytorchLightning-v1.9%20v2.0-blue.svg) | [![Docs](https://img.shields.io/badge/docs-mkdocs%20material-brightgreen)](https://srgan.opensr.eu)<br>![License: Apache](https://img.shields.io/badge/license-Apache%20License%202.0-blue) | [![CI](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml/badge.svg)](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml)<br>[![codecov](https://codecov.io/github/ESAOpenSR/SRGAN/graph/badge.svg?token=LQ69MIMLVE)](https://codecov.io/github/ESAOpenSR/SRGAN) |  [![arXiv](https://img.shields.io/badge/arXiv-2511.10461-b31b1b.svg)](https://arxiv.org/abs/2511.10461)  <br> [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.17590993.svg)](https://doi.org/10.5281/zenodo.17590993)
+| [![PyPI](https://img.shields.io/pypi/v/opensr-srgan)](https://pypi.org/project/opensr-srgan/) | ![PythonVersion](https://img.shields.io/badge/Python-v3.10%20v3.12-blue.svg)<br>![PLVersion](https://img.shields.io/badge/PytorchLightning-v2%2B-blue.svg) | [![Docs](https://img.shields.io/badge/docs-mkdocs%20material-brightgreen)](https://srgan.opensr.eu)<br>![License: Apache](https://img.shields.io/badge/license-Apache%20License%202.0-blue) | [![CI](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml/badge.svg)](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml)<br>[![codecov](https://codecov.io/github/ESAOpenSR/SRGAN/graph/badge.svg?token=LQ69MIMLVE)](https://codecov.io/github/ESAOpenSR/SRGAN) |  [![arXiv](https://img.shields.io/badge/arXiv-2511.10461-b31b1b.svg)](https://arxiv.org/abs/2511.10461)  <br> [![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.17590993.svg)](https://doi.org/10.5281/zenodo.17590993)
 
 
 
@@ -49,7 +49,7 @@ All key knobs are exposed via YAML in the `opensr_srgan/configs` folder:
 * **EMA smoothing:** Enable `Training.EMA.enabled` to keep a shadow copy of the generator. Decay values in the 0.995–0.9999 range balance responsiveness with stability and are swapped in automatically for validation/inference.
 * **Spectral normalization:** Optional for the SRGAN discriminator via `Discriminator.use_spectral_norm` to better control its Lipschitz constant and stabilize adversarial updates. [Miyato et al., 2018](https://arxiv.org/abs/1802.05957)
 * **Wasserstein critic + R1 penalty:** Switch `Training.Losses.adv_loss_type: wasserstein` to enable a critic objective and pair it with the configurable `Training.Losses.r1_gamma` gradient penalty on real images for smoother discriminator updates. [Arjovsky et al., 2017](https://arxiv.org/abs/1701.07875); [Mescheder et al., 2018](https://arxiv.org/abs/1801.04406)
-* **Relativistic average GAN (BCE):** Set `Training.Losses.relativistic_average_d: true` to train D/G on relative real-vs-fake logits instead of absolute logits. This is supported in both Lightning training paths (PL1 and PL2).
+* **Relativistic average GAN (BCE):** Set `Training.Losses.relativistic_average_d: true` to train D/G on relative real-vs-fake logits instead of absolute logits. This is supported in the Lightning 2+ manual-optimization training path.
 The schedule and ramp make training **easier, safer, and more reproducible**.
 
 ---

docs/architecture.md

@@ -2,9 +2,9 @@
 
 This document outlines how ESA OpenSR organises its super-resolution GAN, the major components that make up the model, and how each piece interacts during training and inference.
 
-## Vackground
+## Background
 
-OpenSR-SRGAN follows the single-image super-resolution (SISR) formulation in which the generator learns a mapping from a low-resolution observation $x$ to a plausible high-resolution reconstruction $x'$. The generator head widens the receptive field, a configurable trunk of $N$ residual-style blocks extracts features, and an upsampling tail increases spatial resolution. The residual fusion keeps skip connections active so the network focuses on high-frequency corrections rather than relearning the full signal:
+OpenSR-SRGAN follows the single-image super-resolution (SISR) formulation in which the generator learns a mapping from a low-resolution observation \(x\) to a plausible high-resolution reconstruction \(x'\). The generator head widens the receptive field, a configurable trunk of \(N\) residual-style blocks extracts features, and an upsampling tail increases spatial resolution. The residual fusion keeps skip connections active so the network focuses on high-frequency corrections rather than relearning the full signal:
 $$
 x' = \mathrm{Upsample}\!\left( \mathrm{Conv}_{\text{tail}}\!\left(\mathrm{Body}(x_{\text{head}}) + x_{\text{head}}\right)\! \right).
 $$
@@ -24,9 +24,10 @@ Because every generator variant (residual, RCAB, RRDB, large-kernel attention, E
   total-variation terms. Adversarial supervision uses `torch.nn.BCEWithLogitsLoss` with optional label smoothing.
 * **Optimiser scheduling.** `configure_optimizers()` returns paired Adam optimisers (generator + discriminator) with
   `ReduceLROnPlateau` schedulers that monitor a configurable validation metric.
-* **Training orchestration.** `training_step()` alternates discriminator (`optimizer_idx == 0`) and generator (`optimizer_idx ==
-  1`) updates. During the warm-up period configured by `Training.pretrain_g_only`, discriminator weights are frozen via
-  `on_train_batch_start()` and a dedicated `pretraining_training_step()` computes purely content-driven updates.
+* **Training orchestration.** `setup_lightning()` binds `training_step_PL2()` and enables manual optimisation
+  (`automatic_optimization = False`). Each step performs explicit discriminator and generator optimiser updates; during the
+  warm-up period configured by `Training.pretrain_g_only`, the generator runs content-driven updates while discriminator metrics
+  are logged without stepping discriminator weights.
 * **Validation and logging.** `validation_step()` computes the same content metrics, logs discriminator diagnostics, and pushes
   qualitative image panels to Weights & Biases according to `Logging.num_val_images`.
 * **Inference pipeline.** `predict_step()` automatically normalises Sentinel-2 style 0–10000 inputs, runs the generator,
@@ -53,7 +54,7 @@ The generator zoo lives under `opensr_srgan/model/generators/` and can be select
 * **Stochastic GAN generator (`cgan_generator.py`).** Extends the flexible generator with conditioning inputs and latent noise,
   enabling experiments where auxiliary metadata influences the super-resolution output.
 * **ESRGAN generator (`esrgan.py`).** Implements the RRDBNet trunk introduced with ESRGAN, exposing `n_blocks`, `growth_channels`,
-  and `res_scale` so you can dial in deeper receptive fields and sharper textures. The implementation supports original features like Relativistic Average GAN (RaGAN) and the codebase allows to perform two step training phase (content-oriented pretraining of generator followed by adversarial training with Discriminator) as originally proposed by ESRGAN authors.
+  and `res_scale` so you can dial in deeper receptive fields and sharper textures. The implementation supports original features like Relativistic Average GAN (RaGAN), and the codebase allows a two-step training phase (content-oriented pretraining of the generator followed by adversarial training with the discriminator), as originally proposed by the ESRGAN authors.
 * **Advanced variants (`SRGAN_advanced.py`).** Provides additional block implementations and compatibility aliases exposed in
   `__init__.py` for backwards compatibility.
 

docs/configuration.md

@@ -75,7 +75,7 @@ If you need to reuse the same function for both directions (for example
 | Key | Default | Description |
 | --- | --- | --- |
 | `in_bands` | 6 | Number of input channels expected by the generator and discriminator. |
-| `continue_training` | `False` | Path to a Lightning checkpoint for resuming training (`ckpt_path` on PL ≥ 2, `resume_from_checkpoint` on PL < 2). |
+| `continue_training` | `False` | Path to a Lightning checkpoint for resuming training (`Trainer.fit(..., ckpt_path=...)`). |
 | `load_checkpoint` | `False` | Path to a checkpoint used solely for weight initialisation (no training state restored). |
 
 ## Training
@@ -109,12 +109,13 @@ stable validation imagery. The EMA is fully optional and controlled through the
 | `adv_loss_beta` | `1e-3` | Target weight applied to the adversarial term after ramp-up. |
 | `adv_loss_schedule` | `cosine` | Ramp shape (`linear` or `cosine`). |
 | `adv_loss_type` | `bce` | Adversarial objective (`bce` for classic SRGAN logits, `wasserstein` for a non-saturating critic-style loss). |
-| `relativistic_average_d` | `False` | BCE-only switch for relativistic-average GAN training (real/fake logits are compared against each other's batch mean). Supported in both PL1 and PL2 training-step implementations. |
+| `relativistic_average_d` | `False` | BCE-only switch for relativistic-average GAN training (real/fake logits are compared against each other's batch mean). Supported in the Lightning 2+ manual-optimization training-step implementation. |
 | `r1_gamma` | `0.0` | Strength of the R1 gradient penalty applied to real images (useful with Wasserstein critics). |
 | `l1_weight` | `1.0` | Weight of the pixelwise L1 loss. |
 | `sam_weight` | `0.05` | Weight of the spectral angle mapper loss. |
 | `perceptual_weight` | `0.1` | Weight of the perceptual feature loss. |
 | `perceptual_metric` | `vgg` | Backbone used for perceptual features (`vgg` or `lpips`). |
+| `fixed_idx` | unset | Optional fixed 3-band indices used by perceptual loss when `in_bands > 3` (recommended: `[0, 1, 2]` for RGB+NIR setups). |
 | `tv_weight` | `0.0` | Total variation regularisation strength. |
 | `max_val` | `1.0` | Peak value assumed by PSNR/SSIM computations. |
 | `ssim_win` | `11` | Window size for SSIM metrics. Must be an odd integer. |
@@ -133,6 +134,7 @@ stable validation imagery. The EMA is fully optional and controlled through the
 | `growth_channels` | `32` | ESRGAN-only: growth channels inside each RRDB block. |
 | `res_scale` | `0.2` | Residual scaling used by stochastic/ESRGAN variants. |
 | `out_channels` | `Model.in_bands` | ESRGAN-only: override the number of output bands. |
+| `use_icnr` | `True` | ESRGAN-only: apply ICNR initialization to pre-PixelShuffle convolutions to reduce checkerboard artifacts in low-frequency regions. |
 
 ## Discriminator
 
@@ -200,6 +202,9 @@ The trainer instantiates independent Adam optimisers for the generator and discr
 
 Weight decay exclusions are handled automatically: batch/instance/group-norm layers and bias parameters are filtered into a no-decay group so regularisation only touches convolutional kernels and dense weights. This mirrors best practices for GAN training and keeps normalisation statistics stable.
 
+For ESRGAN runs on multispectral datasets, a practical stability baseline is:
+`Generator.use_icnr: true`, `Training.Losses.fixed_idx: [0, 1, 2]`, and `optim_d_lr <= 0.5 * optim_g_lr`.
+
 ## Schedulers
 
 Both optimisers share the same configuration keys because they use `torch.optim.lr_scheduler.ReduceLROnPlateau`.

docs/index.md

@@ -4,15 +4,15 @@
 
 | **PyPI** | **Versions** | **Docs & License** | **Tests** | **Reference** |
 |:---------:|:-------------:|:------------------:|:----------:|:--------------:|
-| [![PyPI](https://img.shields.io/pypi/v/opensr-srgan)](https://pypi.org/project/opensr-srgan/) | ![PythonVersion](https://img.shields.io/badge/Python-v3.10%20v3.12-blue.svg)<br>![PLVersion](https://img.shields.io/badge/PyTorchLightning-v1.9%20v2.0-blue.svg) | [![Docs](https://img.shields.io/badge/docs-mkdocs%20material-brightgreen)](https://srgan.opensr.eu)<br>![License: Apache](https://img.shields.io/badge/license-Apache%20License%202.0-blue) | [![CI](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml/badge.svg)](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml)<br>[![codecov](https://codecov.io/github/ESAOpenSR/SRGAN/graph/badge.svg?token=LQ69MIMLVE)](https://codecov.io/github/ESAOpenSR/SRGAN) | [![arXiv](https://img.shields.io/badge/arXiv-2511.10461-b31b1b.svg)](https://arxiv.org/abs/2511.10461)<br>[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.17590993.svg)](https://doi.org/10.5281/zenodo.17590993) |
+| [![PyPI](https://img.shields.io/pypi/v/opensr-srgan)](https://pypi.org/project/opensr-srgan/) | ![PythonVersion](https://img.shields.io/badge/Python-v3.10%20v3.12-blue.svg)<br>![PLVersion](https://img.shields.io/badge/PyTorchLightning-v2%2B-blue.svg) | [![Docs](https://img.shields.io/badge/docs-mkdocs%20material-brightgreen)](https://srgan.opensr.eu)<br>![License: Apache](https://img.shields.io/badge/license-Apache%20License%202.0-blue) | [![CI](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml/badge.svg)](https://github.com/ESAOpenSR/SRGAN/actions/workflows/ci.yml)<br>[![codecov](https://codecov.io/github/ESAOpenSR/SRGAN/graph/badge.svg?token=LQ69MIMLVE)](https://codecov.io/github/ESAOpenSR/SRGAN) | [![arXiv](https://img.shields.io/badge/arXiv-2511.10461-b31b1b.svg)](https://arxiv.org/abs/2511.10461)<br>[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.17590993.svg)](https://doi.org/10.5281/zenodo.17590993) |
 
 ![Super-resolved Sentinel-2 example](assets/6band_banner.png)
 
 OpenSR-SRGAN is a comprehensive toolkit for training and evaluating super-resolution GANs on remote-sensing imagery. It packages a flexible generator/discriminator zoo, composable perceptual and reconstruction losses, and the training heuristics
 that make adversarial optimisation tractable—generator warm-up phases, learning-rate scheduling, adversarial-weight ramping, and more. All options are driven by concise YAML configuration files so you can explore new architectures or datasets without
 rewriting pipelines.  
 
-Whether you are reproducing published results, exploring new remote-sensing modalities, or are trying to esablish some benchmarks, OpenSR-SRGAN gives you a clear and extensible foundation for multispectral super-resolution research.
+Whether you are reproducing published results, exploring new remote-sensing modalities, or trying to establish benchmarks, OpenSR-SRGAN gives you a clear and extensible foundation for multispectral super-resolution research.
 
 > This repository and the configs represent the experiences that were made with SR-GAN training for remote sensing imagery. It's neither complete nor claims to perform SOTA SR, but it implements all tweaks and tips that make training SR-GANs easier.
 
@@ -25,7 +25,7 @@ Whether you are reproducing published results, exploring new remote-sensing moda
 * **Battle-tested training loop.** PyTorch Lightning handles mixed precision, gradient accumulation, multi-GPU training, and
   restartable checkpoints while the repo layers in GAN-specific tweaks such as adversarial weight ramping and learning-rate
   restarts.
-* **Lightning 1.x ↔ 2.x compatibility.** The trainer detects the installed Lightning version at runtime and routes to the appropriate automatic- or manual-optimisation step so your configs run unchanged across releases.
+* **Lightning 2+ training path.** Training uses manual optimization (`automatic_optimization=False`) and a single GAN training-step implementation.
 * **Remote-sensing aware defaults.** Normalisation, histogram matching, spectral-band handling, and Sentinel-2 SAFE ingestion are
   ready-made for 10 m and 20 m bands and easily extendable to other sensors.
 
@@ -82,7 +82,7 @@ Whether you are reproducing published results, exploring new remote-sensing moda
 * [Results](results.md) showcases results for some generator/discriminator and dataset combinations.
 
 ## ESA OpenSR
-OpenSR-SRGAN is part of the ESA [OpenSR](https://www.opensr.eu) ecosystem — an open framework for trustworthy super-resolution of multispectral satellite imagery. Within this initiative, this repository serves as the adversarial benchmark suite: it provides standardized GAN architectures, training procedures, and evaluation utilities that complement the other model types implemented in the project (diffusion, transformers, regression) and interfaces with from companion packages such as opensr-utils.
+OpenSR-SRGAN is part of the ESA [OpenSR](https://www.opensr.eu) ecosystem — an open framework for trustworthy super-resolution of multispectral satellite imagery. Within this initiative, this repository serves as the adversarial benchmark suite: it provides standardized GAN architectures, training procedures, and evaluation utilities that complement the other model types implemented in the project (diffusion, transformers, regression), and it interfaces with companion packages such as opensr-utils.
 
 ## Citation