sRender

sRender ends the Shape portion of a Flow. Upstream tools such as sRectangle, sEllipse, sPolygon, sNGon, sStar, sText, sMerge, sBoolean, sDuplicate, sGrid and sOutline continue to pass vector Shape data; downstream of sRender the stream is a normal 2D image.

Blackmagic documents Input1 as an orange required Shape input. It accepts the output of the final Shape node or compound Shape branch. With no Input1 there is no Shape geometry to rasterize; the optional Effect Mask is not a substitute source.

sRender also exposes an optional blue Effect Mask. A Rectangle, Polygon, B-Spline, Bitmap or other mask can gate where the rendered result is visible without becoming part of the incoming Shape geometry.

That topology is essential: Shape data, mask data and raster image data are separate domains. Shape goes to Input1, a mask goes to Effect Mask, and Image Output feeds ordinary Glow, Blur, Merge, Transform, Color Corrector or MediaOut nodes.

Global In/Out defines the active frame range. The captured Inspector shows 0–119 in a 120-frame composition; that is screenshot evidence, not a universal factory default.

Process Mode selects field processing. The captured menu shows Full Frames, NTSC Fields, PAL/HD Fields and reversed field-order variants. Full Frames is the appropriate baseline for normal progressive compositions.

Field modes are for genuine interlaced/field-based delivery. They are not quality or smoothness switches and should not be enabled in a progressive comp without a field workflow reason.

Width and Height define the raster canvas in pixels when Auto Resolution is disabled. They do not edit the vector primitive the way Width/Height on sRectangle do; they define the image into which the Shape stream is rasterized.

Pixel Aspect X and Y describe the raster pixel shape. A square-pixel workflow uses 1.0 / 1.0. A wrong pixel aspect can visually stretch correctly authored Shape geometry.

According to the Shape reference, Auto Resolution locks Width, Height and Pixel Aspect to the composition Frame Format. This is more specific than saying the node simply 'chooses a size automatically'; the incoming Shape stream has no ordinary raster dimensions to inherit.

With Auto Resolution disabled, sRender can act as an independent rasterizer with an explicit canvas—for example, a 2048×2048 graphics element inside a larger composition. In that case Width/Height/Pixel Aspect become part of the pipeline contract.

Depth controls the precision of the created image. The captured Resolve 21 menu shows Default, int8, int16, float16 and float32. Higher precision costs more memory/bandwidth but provides more numerical accuracy and HDR headroom.

int8 is suitable only for simple SDR cases where downstream processing is limited. For heavy Glow/Blur/color math and HDR or linear work, float16 is usually a safer balance; float32 is reserved for cases that truly need the extra precision.

Source Color Space should not be described as a full gamut conversion. Blackmagic describes this block as assigning/passing source color-space metadata; Space lets you explicitly identify a color space for downstream tools to use.

Color Space Type Auto reads/passes available metadata, while Space reveals an explicit color-space choice. Use a proper color-management/Gamut/CST workflow when actual pixel conversion between gamuts is required.

Source Gamma Space similarly identifies or reads gamma metadata through Curve Type. Auto uses metadata, Space exposes a gamma-space choice, and Log exposes log/linear controls.

Remove Curve changes pixel values: it removes the interpreted gamma/log curve to produce a linearized output. It is therefore not a harmless tagging option and should not be enabled by habit.

Pre-Divide / Post-Multiply addresses premultiplied alpha around color/gamma processing. RGB is divided by alpha before processing and multiplied by alpha again afterward, helping protect translucent edges from fringe values.

Pre-Divide / Post-Multiply is not a universal halo-removal switch. First confirm that a premultiplied-alpha color/gamma operation is actually causing the edge problem.

Rasterization is the point where direct Shape editing ends. After sRender, Shape-only tools such as sBoolean, sGrid or sDuplicate no longer receive Shape data, so construction, repetition, boolean work and Shape-native styling normally belong upstream.

After sRender, ordinary image effects become appropriate. A classic production chain is Shape generators/modifiers → sRender → Soft Glow/Blur/Color → Merge → MediaOut.

Effect Mask is useful when a complex compound Shape should be spatially restricted at the final raster boundary without rebuilding the vector branch. If the restriction must participate in Shape geometry or boolean logic, solve it upstream in the Shape system instead.

Perform raster-quality QC at 100% Viewer scale. Thin outlines, rounded caps, subpixel motion and anti-aliasing can look deceptively similar in Fit-to-view while differing at target raster resolution.

Inspect alpha separately from RGB. sRender alpha reflects authored Shape style/alpha, upstream overlap behavior and the optional Effect Mask; downstream Glow or Blur can then expand RGB and/or alpha footprint.

When output dimensions look wrong, inspect Auto Resolution and Frame Format first, then Width/Height/Pixel Aspect. Do not compensate for a raster-format error by distorting the Shape geometry itself.

When brightness changes unexpectedly after Remove Curve, A/B Remove Curve first, then verify Curve Type and downstream color management. Source Color Space metadata and gamma removal are different layers of the pipeline.

Production QC for sRender should isolate topology, active frame range, progressive/field mode, raster size, pixel aspect, depth, alpha edge and color/gamma metadata so one setting does not hide another.