diff --git a/test/browser/browser.sh b/test/browser/browser.sh
index a40c7ed37..0d5c581b2 100755
--- a/test/browser/browser.sh
+++ b/test/browser/browser.sh
@@ -74,12 +74,8 @@ for retry in $(seq 5); do
     sleep $((5 * retry * retry))
 done
 
-# copy images for user podman tests; podman insists on user session
-loginctl enable-linger $(id -u admin)
-for img in localhost/test-alpine localhost/test-busybox localhost/test-registry; do
-    podman save  $img | sudo -i -u admin podman load
-done
-loginctl disable-linger $(id -u admin)
+# image setup, shared with upstream tests
+$TESTS/../vm.install
 
 systemctl enable --now cockpit.socket podman.socket
 
diff --git a/test/check-application b/test/check-application
index 9d790b2a5..d7cf55313 100755
--- a/test/check-application
+++ b/test/check-application
@@ -140,11 +140,19 @@ class TestApplication(testlib.MachineCase):
                                                     ssh_port=m.ssh_port,
                                                     identity_file=m.identity_file)
 
-        # Enable user service as well
-        self.admin_s.execute("systemctl --user stop podman.service; systemctl --now --user enable podman.socket")
-        self.restore_dir("/home/admin/.local/share/containers")
-        self.addCleanup(self.admin_s.execute, "systemctl --user stop podman.service podman.socket || true")
-        # Ubuntu 22.04 has old podman that does not know about --time
+        # Enable user service as well; copy our images (except cockpit/ws) from system
+        self.admin_s.execute("""
+            systemctl --user stop podman.service
+            for img in $(ls /var/lib/test-images/*.tar | grep -v cockpitws); do podman load < "$img"; done
+            systemctl --now --user enable podman.socket
+            """)
+        self.addCleanup(self.admin_s.execute, """
+            systemctl --user stop podman.service podman.socket
+            podman system reset --force
+            """)
+        # HACK: system reset has 10s timeout, make that faster with an extra `stop`
+        # https://github.com/containers/podman/issues/21874
+        # Ubuntu 22.04 has old podman that does not know about rm --time
         if m.image == 'ubuntu-2204':
             self.addCleanup(self.admin_s.execute, "podman rm --force --all", timeout=300)
             self.addCleanup(self.admin_s.execute, "podman pod rm --force --all", timeout=300)
diff --git a/test/vm.install b/test/vm.install
index eb2cd9890..f6c52cf53 100755
--- a/test/vm.install
+++ b/test/vm.install
@@ -26,13 +26,12 @@ fi
 # Since 4.0 podman now ships the pause image
 podman images --format '{{.Repository}}:{{.Tag}}' | grep -Ev 'localhost/test-|pause|cockpit/ws' | xargs -r podman rmi -f
 
-# copy images for user podman tests; podman insists on user session
-loginctl enable-linger $(id -u admin)
-images=$(podman images --format '{{.Repository}}:{{.Tag}}')
-for img in $images; do
-	podman save  $img | sudo -i -u admin podman load
+# tests reset podman, save the images
+mkdir -p /var/lib/test-images
+for img in $(podman images --format '{{.Repository}}:{{.Tag}}'); do
+    fname="$(echo "$img" | tr -dc '[a-zA-Z-]')"
+    podman save -o "/var/lib/test-images/${fname}.tar" "$img"
 done
-	loginctl disable-linger $(id -u admin)
 
 # 15minutes after boot tmp files are removed and podman stores some tmp lock files
 systemctl disable --now systemd-tmpfiles-clean.timer