⚡ How to make qwen work: https://tensor.art/articles/906954308069633221
A telephoto lens is a camera lens with a long focal length (typically 70mm to 600mm or more). Its main characteristic is bringing distant subjects visually closer while compressing perspective.
✨ Why It Matters in Your Prompts
When you say telephoto, you’re telling the model to:
Avoid wide-angle distortion (no exaggerated foreground, no fisheye).
Generate grand, layered imagery (background closer, subject more integrated).
Emphasize clarity and monumentality (everything feels heavy, historical, or cosmic).
🔑 Prompting Scheme Overview
We’ve been building prompts in two layers:
Positive Prompt → describes what we want, in rich cinematic terms.
Negative Prompt → filters out everything we don’t want (style mismatches, distortions, anachronisms, clutter).
Additionally, we make a Compact Version (one-liner positives + negatives) for quick use or for models that respond better to shorter inputs.
🧩 Positive Prompt Strategy
The positive prompt is structured in 4 steps:
Camera/Style Definition
Start with framing, e.g. “telephoto shot, archival-style, astrophotograph, cinematic composition”.
This sets the photographic language (lens type, focus style, mood).
Subject Definition
Describe the main subject(s) clearly: “Roman soldier from behind,” “angel guiding silhouettes,” “Los Tercios with pikes.”
This grounds the image in concrete, recognizable forms.
Scene/Environment
Add the setting and context: “pyramids in the desert,” “burning Sodom and Gomorrah in the distance,” “icy glaciers in Greenland.”
Use compressed perspective keywords to emphasize telephoto style.
Color/Atmosphere/Meaning
Final layer adds emotional tone and color palette: “vivid sapphire blues and turquoise,” “golden light of destiny,” “biblical, reverent, monumental.”
This creates impact and narrative significance.
🚫 Negative Prompt Strategy
The negative prompt works as a safeguard. It usually covers:
Technical flaws
lowres, blurry, noise, artifacts, motion blur, fisheye, wide-angle distortion, shallow DOF.
Stylistic mismatches
cartoon, anime, CGI, 3D render, sketch, surreal fantasy, neon glow, oversaturation, HDR glow.
Anachronisms/Intrusions
cars, planes, modern weapons, tourists, skyscrapers, crowds, fantasy creatures.
This ensures the scene stays consistent to its intended period or theme.
Generation artifacts
tiling, repeating patterns, glitches, watermark, text, logos.
⚡ Compact Version Use
The compact prompts are distilled versions:
Positive compact → one descriptive sentence with key camera/subject/atmosphere words.
Negative compact → one short filter list with only the most disruptive issues.
These are useful when:
The model prefers brevity.
You want to chain prompts quickly for variations.
You need base templates for experimentation.