HaanIlango commited on
Commit
6b32365
路
verified 路
1 Parent(s): 8fc1ba3

Update app.py

Browse files
Files changed (1) hide show
  1. app.py +14 -13
app.py CHANGED
@@ -761,10 +761,10 @@ def analyze(pil_image, overlay_strength):
761
 
762
  summary += (
763
  "\n\n---\n"
764
- "**This tool flags what's off. A designer knows what's right.** I'm Haan - "
765
- "a brand identity and visual designer with 12+ years turning insights like "
766
- "these into designs that actually convert, stay true to your brand, and "
767
- "look intentional (not just \"fixed\").\n\n"
768
  "[Visit haanilango.com](https://www.haanilango.com) 路 Tel: +65 9027 2070"
769
  )
770
 
@@ -780,23 +780,24 @@ with gr.Blocks(title="Design Analyzer") as demo:
780
  "**by Haan Ilango** 路 [haanilango.com](https://www.haanilango.com/) 路 Tel: +65 9027 2070"
781
  )
782
  gr.Markdown(
783
- "**This tool helps you improve your design - it's helped me improve mine too.**\n\n"
784
- "I built it because design decisions shouldn't come down to whose opinion carries "
785
- "more weight in the room. Good design sometimes loses to confident opinions instead "
786
- "of evidence. This gives you something concrete to point to - so the conversation "
787
- "shifts from \"I just don't like it\" to a case backed by evidence, not just opinion.\n\n"
788
- "If you'd like it done professionally, I'm here for that too."
 
 
789
  )
790
  gr.Markdown("Upload a design to check where attention is drawn, plus text contrast and readability.")
791
  gr.Markdown(
792
  "### Why this tool is different\n\n"
793
  "Most AI attention-prediction tools quote a big accuracy number "
794
- "(\"up to 96% accurate!\") with no published methodology behind it. "
795
- "We'd rather show our work.\n\n"
796
  "- **Built on a real, peer-reviewed model** (Kroner et al., 2020) - "
797
  "not a black box. See the citation below.\n"
798
  "- **Honest about uncertainty.** If text can't be reliably read, "
799
- "we say so instead of faking a score.\n"
800
  "- **Works on any static visual** - posters, social posts, app "
801
  "screens, banner ads, print, packaging.\n"
802
  "- **Attention + readability in one pass**, not two separate tools.\n"
 
761
 
762
  summary += (
763
  "\n\n---\n"
764
+ "**This tool flags what's off. A designer knows what's right.** "
765
+ "Haan is a brand identity and visual designer with 12+ years turning "
766
+ "insights like these into designs that actually convert, stay true to "
767
+ "your brand, and look intentional (not just \"fixed\").\n\n"
768
  "[Visit haanilango.com](https://www.haanilango.com) 路 Tel: +65 9027 2070"
769
  )
770
 
 
780
  "**by Haan Ilango** 路 [haanilango.com](https://www.haanilango.com/) 路 Tel: +65 9027 2070"
781
  )
782
  gr.Markdown(
783
+ "**This tool helps improve design - grounded in evidence, not opinion.**\n\n"
784
+ "Design decisions shouldn't come down to whose opinion carries the most "
785
+ "weight in the room. Good design sometimes loses to confident opinions "
786
+ "instead of evidence. This gives you something concrete to point to - "
787
+ "so the conversation shifts from \"I just don't like it\" to a case "
788
+ "backed by evidence, not just opinion.\n\n"
789
+ "Professional design services are also available for those who'd "
790
+ "like it done for them."
791
  )
792
  gr.Markdown("Upload a design to check where attention is drawn, plus text contrast and readability.")
793
  gr.Markdown(
794
  "### Why this tool is different\n\n"
795
  "Most AI attention-prediction tools quote a big accuracy number "
796
+ "(\"up to 96% accurate!\") with no published methodology behind it.\n\n"
 
797
  "- **Built on a real, peer-reviewed model** (Kroner et al., 2020) - "
798
  "not a black box. See the citation below.\n"
799
  "- **Honest about uncertainty.** If text can't be reliably read, "
800
+ "that's flagged instead of faking a score.\n"
801
  "- **Works on any static visual** - posters, social posts, app "
802
  "screens, banner ads, print, packaging.\n"
803
  "- **Attention + readability in one pass**, not two separate tools.\n"