图片｜Grok Imagine Image Pro 生图

This is the same node as the pack's standard Grok Imagine text-to-image - identical inputs, identical workflow - with one change: the model is grok-imagine-image-pro instead of grok-imagine-image, and Tikpan prices it at 0.728 RMB per image versus 0.208 for the standard. About 3.5x the price for what is, honestly, the last 20% of quality. If you're making social-media filler or testing prompts, buy the standard node. If the image is going into a paid ad, a product listing, or anywhere a client's eyes will land, Pro's extra detail and more believable lighting are usually worth the difference.

Everything else about it is the pack's plain text-to-image node, so the mechanics won't surprise you if you've used its sibling. Prompt in via 生成指令, POST to Tikpan's /v1/images/generations, image tensor out. There's no editing input here - no reference image port - this one is strictly "make me a new image from text." The edit flavor of Pro lives in a separate node.

The inputs that matter

• API_密钥 - your sk- key from https://tikpan.com. Prepaid, no free tier.
• 生成指令 - the prompt; specific English wins. The default is a cinematic product poster, which is this pack's target audience in one line.
• 清晰度 - 1k vs 2k. On Pro, the 2k bump is where the money shows: more texture in materials, sharper fine detail. Still, test at 1k first, because Pro at 1k beats standard at 2k for most product work and costs about the same.
• 画面比例 - auto or the usual 1:1 / 16:9 / 9:16.
• 生成张数 - 1 to 10, each billed.

The outputs are Image (IMAGE) and Log (STRING), exactly like the standard node - wire Image into a Save/Preview node and you're done.

The Pro-versus-standard call

Here's the honest rule of thumb. Standard Grok Imagine is a workhorse: fast, cheap, good enough for content fill. Pro is for the deliverables. Where people get burned is treating Pro as a drop-in upgrade for a workflow that was working - you pay 3.5x for gains you only notice at 2k on a big screen. My advice: keep standard in the draft stage, reach for Pro only when a prompt is already good and the image actually matters. The one-line test - "would I notice this difference on a phone?" - answers it for most people most of the time.

Install and first run

Installation is the pack's standard affair, and there are no extra dependencies beyond what ComfyUI ships:

cd ComfyUI/custom_nodes
git clone https://github.com/htrert/ComfyUI-Tikpan-Pro

Restart, or search "Tikpan" in ComfyUI Manager. Paste your key, write a prompt, run. First result comes back in a few seconds - it's a synchronous API call, so the node blocks until the image is done.

Common issues

Same upstream error family as every node here: 401 bad key, 402 empty balance, 429 too many requests (slow down, wait 30–60s). Because Pro costs more per call, double-check your prompt before submitting and keep an eye on the Log output for whether a failed request still drew from your balance - a failed call at Pro prices stings more. And if you're batch-generating, 跳过错误 will return a black image on failure instead of aborting the whole run.