🔳 CR Aspect Ratio Social Media

Same machinery as CR Aspect Ratio, different preset list. CR Aspect Ratio Social Media ships with the dimensions the platforms actually want - Instagram Portrait (1080x1350), Instagram Square, Stories/Reels (1080x1920), Facebook, TikTok (1080x1920), YouTube Banner (2560x1440), LinkedIn profile and page covers, and a handful more. Pick the platform, get width, height, and a matching empty latent. It's for when you're generating for a specific feed and don't feel like Googling "what size is a LinkedIn banner" for the tenth time.

How it works

Identical to its general-purpose sibling. You select an aspect_ratio from the social-media presets, and the node sets width and height and builds an empty latent at that size. custom falls back to your typed width/height. swap_dimensions flips orientation. prescale_factor and upscale_factor are pass-through values for a hires pipeline - they don't resize on their own, they're numbers you route to the nodes that do.

Here's the honest caveat, and it's a real one: these are canvas sizes, not diffusion-friendly sizes. A YouTube banner at 2560x1440 or a LinkedIn cover at 1584x396 is nowhere near the resolution or aspect ratio SD1.5 or SDXL was trained on. Generate a raw diffusion image straight into a 4:1 banner and it'll look stretched and repetitive - the model doesn't know how to fill a shape that extreme. The sane workflow is to generate at a native resolution and then compose or outpaint onto the social canvas, or use the square/portrait presets (Instagram's 1080x1080 and 1080x1350 are close enough to SDXL-native to work directly). Treat the wide banner presets as target dimensions for a later step, not as your sampling resolution.

The inputs and outputs that matter

• aspect_ratio - the platform preset. The reason the node exists.
• swap_dimensions (Off/On) - flip orientation.
• batch_size - number of latents.
• width / height - used when aspect_ratio is custom.

Outputs:

• empty_latent (LATENT) - into your KSampler. But re-read the caveat before you sample straight into a banner size.
• width / height (INT) - useful as the target for a resize or outpaint step.
• upscale_factor / prescale_factor (FLOAT), batch_size (INT) - pass-throughs.
• show_help - wiki link, ignore.

How to install it

Comfyroll Studio (Suzie1 and RockOfFire), a mature utility pack - no models, no heavy dependencies.

• ComfyUI Manager - search Comfyroll Studio, install, restart.
• Manual - cd ComfyUI/custom_nodes && git clone https://github.com/Suzie1/ComfyUI_Comfyroll_CustomNodes.git, then restart.

Also on CivitAI.

Common issues

The one that surprises people is above: extreme social banner ratios produce ugly diffusion output because they're miles from any model's training resolution. Use them as final canvas targets, generate at native sizes, and compose.

If the node's missing entirely, that's the pack failing to load - Comfyroll Studio: Failed to load Graphics nodes in the log, ending in NameError: name 'CR_HalftoneGrid' is not defined. That NameError is a symptom of a real import error higher up, typically an incompatible Pillow version from another custom node, and it can take the whole pack offline. Reinstall Pillow (pip install --upgrade --force-reinstall pillow) into ComfyUI's Python and restart.