Revising NVENC settings for recording and streaming; Adding info about algorithms on recipe page; Adding tmix script

docs/video/ffmpeg/misc-scripts.md

@@ -147,6 +147,8 @@ summary {
 
     See Frost's audio scripts: https://github.com/Thqrn/ffmpeg-scripts/
 ??? quote "tmix"
+
+    https://github.com/Thqrn/ffmpeg-scripts/blob/main/video%20resampler.bat
 ??? quote "Cut from point a to b"
 
     You can use [`-ss` and `-to`](https://ffmpeg.org/ffmpeg-all.html#Main-options), but you should really use a [cutter](../cutters/index.md)

docs/video/obs/output.md

@@ -31,7 +31,7 @@ It's recommended to use the [latest version of OBS](https://github.com/obsprojec
 
         Fragmented MP4 is notorious to mess up in VEGAS, though remuxing to regular MP4 fixes it. 
 
-    In general, encoding speed fastest to slowest is NVENC (NVIDIA), AMF (AMD), QuickSync (Intel iGPU), then x264/5 (CPU)
+    In general, encoding speed from fastest to slowest is NVENC (NVIDIA), AMF (AMD), QuickSync (Intel iGPU), then x264/5 (CPU)
 
     Regarding `CQP` and `CFR` [rate controls](../codecguide.md#rate-controls): It's inverted compared to CBR, 0 is lossless and 51 is most compressed
 
@@ -40,38 +40,56 @@ It's recommended to use the [latest version of OBS](https://github.com/obsprojec
 
     === ":simple-nvidia: (NVIDIA) NVENC"
 
-        !!! abstract "High FPS recording configuration"
+        !!! abstract "Note for File Size Efficiency"
 
-            If you want to record in 120+FPS you're advised to configure the following:
+            The settings provided here are specifically for high framerate recoring, which prefer performance over low file sizes. If file sizes are an issue, consider making the following changes:
 
-            * <div class="annotate" markdown>`Preset`: P1 - P4(1)</div>
+            * <div class="annotate" markdown>`Preset`: P2 - P7(1)</div>
 
-                1. OBS describes low Presets as "Lower Quality" in the context of <u>streaming with CBR in bandwidth-limited scenarios</u>, if used for recording it'll be much more performant to record at the cost of bigger file sizes.
+                1. This setting does not impact quality when using CQP, only the efficiency of quality to file size. Quality is only impacted when using contrained-bitrate rate control modes such as CBR or VBR, making it ideal for streaming.
 
-            * `Multipass Mode`: Single pass
-            * `Look-ahead`: Unchecked
-            * `Psycho Visual Tuning`: Unchecked
+            * `Multipass Mode`: Two Passes (Quarter Resolution)
+            * `Look-ahead`: Checked
+            * `Psycho Visual Tuning`: Checked
             * `Max B-frames`: 0
-            * `Keyframe Interval`: 0 (auto)
+            * `Keyframe Interval`: 0 s (auto)
 
-            :   I've had the habit of turning these off / setting them to 0 for high FPS recording, they're probably relevant for streaming / high efficiency recording though.
+            :   Adjust as needed to achieve desired file size without encoding lag.
 
 
+        * `Rescale Output`: <kbd>Disabled</kbd>
+
         * `Rate Control`: <kbd>CQP</kbd> / <kbd>CQ Level</kbd> / <kbd>Constant QP</kbd>
 
-        :   Much more adaptive than CBR, which always spits out the same constant bitrate.
+        :   Minimal performance impact without compromising on quality since CQP does not target a certain bitrate like CBR/VBR do.
 
         * `CQ Level`: <kbd>18</kbd>
 
-        * `Preset`: <kbd>P7: Slowest (Best Quality)</kbd>
+        :   Can be slightly raised or lowered, but 18 is the best balance between visually lossless and huge files. There is essentially no point in going below it, as quality is not measurably increased.
+
+        * `Keyframe interval`: <kbd>0 s</kbd>
+
+        * `Preset`: <kbd>P1: Slowest (Best Quality)</kbd>
+
+        :   Does not impact quality when using CQP. Higher values give smaller file sizes at the cost of performance.
+
+        * `Tuning`: <kbd>High Quality</kbd>
+
+        * `Multipass Mode`: <kbd>Single Pass</kbd>
+
+        :   Using any other option massively reduces performance.
 
-        * `Multipass Mode`: <kbd>Two Passes (Full Resolution)</kbd>
+        * `Profile`: <kbd>baseline</kbd>
 
-        :   These two settings considerably affect the encoding speed and efficiency, they've been recently introduced to OBS and I'm not that knowledgeable regarding them.
+        :   Slightly increased performance over high and main at the cost of larger files.
 
-        * `GPU`: 0
+        * `Look-ahead`: Unchecked; `Adaptive Quantization`: Unchecked
 
-        :   This is the GPU index, only relevant to mess with if you have multiple GPUs
+        :   Only useful when targeting a bitrate (VBR, CBR), but huge performance impacts. Unnecessary for CQP.
+
+        * `B-Frames`: <kbd>0</kbd>
+
+        :   Automatic, though they're only even used if you're not using the baseline profile.
 
     === ":custom-amd: (AMD) AMF"
 
@@ -154,17 +172,24 @@ It's recommended to use the [latest version of OBS](https://github.com/obsprojec
 
         * `Rate Control`: <kbd>CBR</kbd>
 
-        * `Bitrate`: Depends on platform and upload bandwidth
-
-        :   See links above
+        * `Bitrate`: Depends on platform and upload bandwidth, typically between 8000-1000 kbps but see links above.
 
         * `Preset`: <kbd>P7: Slowest (Best Quality)</kbd>
 
         :   That'll provide the most efficient encoding for the given bandwidth
 
-        * `Multipass Mode`: <kbd>Two Passes (Full Resolution)</kbd>
+        * `Multipass Mode`: <kbd>Two Passes (Quarter Resolution)</kbd>
+             ///////////////////////
+
+        * `Keyframe interval`: <kbd>0 s</kbd>
+
+        * `Tuning`: <kbd>High Quality</kbd>
+
+        * `Profile`: <kbd>high</kbd>
+
+        * `Look-ahead`: Checked; `Adaptive Quantization`: Checked
 
-        :   These two settings considerably affect the encoding speed and efficiency, they've been recently introduced to OBS and I'm not that knowledgeable regarding them.
+        * `B-Frames`: <kbd>4</kbd>
 
         <iframe width="688" height="387" src="https://www.youtube-nocookie.com/embed/uAqLJ3sxudU?color=white" frameborder=0 allowfullscreen></iframe>
 

docs/video/smoothie/recipe.md

@@ -147,11 +147,16 @@ Also see [pre-interp](#pre-interp), a slower, more accurate interpolation method
 
 :   Sets the algorithm. Taken from same docs:
 
-    * `2` - This makes strong predictions which can be useful for cartoons but can also leave big artifacts.
-    * `13` - This is the most intelligent algorithm since it masks many artifacts, but it is not as smooth as 23.
-    * `23` - This is the smoothest algorithm, but it does not have the artifact masking that 13 has.
+
+
+    * `1` - Not recommended. Sharp picture, no blending, but moves pixels from previous frame to next using only data from previous frames.
+    * `2` - Only recommended for 2D animation. Like 1 but uses the next frame too.
+    * `11` - Almost as good as 13; uses a time-weighted blend of both forward and backward vectors.
+    * `13` - Best option for low framerate input. 11 but with a dynamic median. Minimal artifacting, but halos can appear around moving objects.
+    * `21` - 11 but with masking, which can reduce halos.
+    * `23` - "Smoothest" but can cause slight artifacts. 21 but with extra data from adjacent frames.
 
-    Most people use 23 / 13
+    Most people use 23 / 13.
 
 `block size`: auto