diff --git a/pcsx2/GS/GSCapture.cpp b/pcsx2/GS/GSCapture.cpp
index 057ba34a8c030..367bc3f769b60 100644
--- a/pcsx2/GS/GSCapture.cpp
+++ b/pcsx2/GS/GSCapture.cpp
@@ -460,6 +460,8 @@ bool GSCapture::BeginCapture(float fps, GSVector2i recommendedResolution, float
 				}
 			}
 		}
+		if (sw_pix_fmt == AV_PIX_FMT_VAAPI)
+			sw_pix_fmt = AV_PIX_FMT_NV12;
 		s_video_codec_context->pix_fmt = sw_pix_fmt;
 
 		// Can we use hardware encoding?