diff --git a/hippunfold/workflow/rules/nnunet.smk b/hippunfold/workflow/rules/nnunet.smk
index a86a6623..ab21a4b0 100644
--- a/hippunfold/workflow/rules/nnunet.smk
+++ b/hippunfold/workflow/rules/nnunet.smk
@@ -2,32 +2,35 @@ import re
 
 
 def get_nnunet_input(wildcards):
-    if config["modality"] == "T2w":
-        nii = (
-            bids(
-                root=work,
-                datatype="anat",
-                **config["subj_wildcards"],
-                suffix="T2w.nii.gz",
-                space="corobl",
-                desc="preproc",
-                hemi="{hemi}",
-            ),
-        )
+    T1w_nii = bids(
+        root=work,
+        datatype="anat",
+        **config["subj_wildcards"],
+        suffix="T1w.nii.gz",
+        space="corobl",
+        desc="preproc",
+        hemi="{hemi}",
+    )
+    T2w_nii = bids(
+        root=work,
+        datatype="anat",
+        **config["subj_wildcards"],
+        suffix="T2w.nii.gz",
+        space="corobl",
+        desc="preproc",
+        hemi="{hemi}",
+    )
+    if (config["modality"] == "T1w" or config["modality"] == "T2w") and config[
+        "force_nnunet_model"
+    ] == "T1T2w":
+        return (T1w_nii, T2w_nii)
+
+    elif config["modality"] == "T2w":
+        return T2w_nii
     elif config["modality"] == "T1w":
-        nii = (
-            bids(
-                root=work,
-                datatype="anat",
-                **config["subj_wildcards"],
-                suffix="T1w.nii.gz",
-                space="corobl",
-                desc="preproc",
-                hemi="{hemi}",
-            ),
-        )
+        return T1w_nii
     elif config["modality"] == "hippb500":
-        nii = bids(
+        return bids(
             root=work,
             datatype="dwi",
             hemi="{hemi}",
@@ -37,7 +40,6 @@ def get_nnunet_input(wildcards):
         )
     else:
         raise ValueError("modality not supported for nnunet!")
-    return nii
 
 
 def get_model_tar():
@@ -94,6 +96,19 @@ def parse_trainer_from_tar(wildcards, input):
     return trainer
 
 
+def get_cmd_copy_inputs(wildcards, input):
+    in_img = input.in_img
+    if isinstance(in_img, str):
+        # we have one input image
+        return f"cp {in_img} tempimg/temp_0000.nii.gz"
+    else:
+        cmd = []
+        # we have multiple input images
+        for i, img in enumerate(input.in_img):
+            cmd.append(f"cp {img} tempimg/temp_{i:04d}.nii.gz")
+        return " && ".join(cmd)
+
+
 rule run_inference:
     """ This rule uses either GPU or CPU .
     It also runs in an isolated folder (shadow), with symlinks to inputs in that folder, copying over outputs once complete, so temp files are not retained"""
@@ -101,7 +116,7 @@ rule run_inference:
         in_img=get_nnunet_input,
         model_tar=get_model_tar(),
     params:
-        temp_img="tempimg/temp_0000.nii.gz",
+        cmd_copy_inputs=get_cmd_copy_inputs,
         temp_lbl="templbl/temp.nii.gz",
         model_dir="tempmodel",
         in_folder="tempimg",
@@ -148,7 +163,7 @@ rule run_inference:
         # run inference
         #copy from temp output folder to final output
         "mkdir -p {params.model_dir} {params.in_folder} {params.out_folder} && "
-        "cp {input.in_img} {params.temp_img} && "
+        "{params.cmd_copy_inputs} && "
         "tar -xf {input.model_tar} -C {params.model_dir} && "
         "export RESULTS_FOLDER={params.model_dir} && "
         "export nnUNet_n_proc_DA={threads} && "