APNext Time

Want to place your image in "ancient Egypt at dusk" or "the 1980s at golden hour" without hunting for the exact phrasing? APNext Time is a dropdown for the when. It offers time of day, decades, centuries, and historical eras, and it appends whatever you pick to your prompt. It's one of the dagthomas category builders - the pack the community calls the SDXL Auto Prompter - and it's the smallest, tightest one of the bunch, which honestly makes it the most reliably useful.

Time and era terms tend to hit hard in image models because they pull whole aesthetics with them - say "1920s" and you get the fashion, the film grain, the color palette for free. That's why this node punches above its size.

How it works

No LLM, no key, nothing downloads. The node carries curated lists baked into the pack - 33 time-of-day phrases, 14 decades, 15 centuries, 90 historical eras - and each dropdown gives you None, a specific pick, Random, or Multiple Random. On run it collects the resolved values, joins them with your separator, and appends them to your base prompt. Rolls are seeded, so a combination you like is reproducible.

The inputs and outputs that matter

• prompt (required, multiline) - your base text; the picks land after it.
• separator (default ,) - the joiner between terms.
• The four dropdowns - time (dawn, dusk, golden hour…), decades, century, eras - each None, a specific pick, Random, or Multiple Random.
• seed - fixes the random rolls.
• attributes (boolean, default off) - adds extra descriptive words where the pack's data attaches them.
• string (optional) - extra text to merge.

Two STRING outputs: prompt (the assembled text) and random (the resolved-random version, so you can reuse the exact roll). Wire either into a CLIP Text Encode.

A small practical note: time (dusk, golden hour, blue hour) and eras/decades do different jobs - one sets lighting, the other sets the whole period aesthetic. Combining one of each is the sweet spot; both usually cooperate rather than fight.

How to install it

ComfyUI Manager: search comfyui_dagthomas or "SDXL Auto Prompter", install, restart. Or manually:

cd ComfyUI/custom_nodes
git clone https://github.com/dagthomas/comfyui_dagthomas
cd comfyui_dagthomas
pip install -r requirements.txt

Restart ComfyUI. The requirements list is large (transformers, decord, provider SDKs) because the pack also ships LLM and vision nodes - Time needs none of that and runs entirely locally.

Common issues & troubleshooting

Time is low-drama because its lists are focused, but two things are worth knowing. First, an era term steers style, so it can quietly override other cues - ask for "ancient Egypt" and a modern subject may pick up period dress and materials you didn't ask for. That's usually the point, but it's why contradictory combos ("1850s, cyberpunk") average out to mush.

Second, the pack-wide caveat: the SDXL Auto Prompter name is honest. This emits comma-separated tags, which suit CLIP-encoded models (SDXL, Illustrious, Pony). On 2026 LLM-encoded models (Flux, Z-Image, Anima) the encoder reads a prompt as an instruction, so a period phrase folded into a sentence works better than a comma-tag dump - and dumping Multiple Random across every category buries your subject past the ~75-100-token point where those models start to drift. Time is small enough that this rarely bites, but keep it to one or two picks.

Finally, keep the pack updated - it's large and actively developed, and a stale copy is the usual cause of a load error. A git pull or Manager update has historically been the fix.